From owner-freebsd-performance@FreeBSD.ORG Sun Jan 7 06:48:05 2007 Return-Path: X-Original-To: performance@freebsd.org Delivered-To: freebsd-performance@FreeBSD.ORG Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id 044FE16A403; Sun, 7 Jan 2007 06:48:05 +0000 (UTC) (envelope-from gnn@neville-neil.com) Received: from mrout1-b.corp.dcn.yahoo.com (mrout1-b.corp.dcn.yahoo.com [216.109.112.27]) by mx1.freebsd.org (Postfix) with ESMTP id A956613C428; Sun, 7 Jan 2007 06:48:04 +0000 (UTC) (envelope-from gnn@neville-neil.com) Received: from minion.local.neville-neil.com (proxy8.corp.yahoo.com [216.145.48.13]) by mrout1-b.corp.dcn.yahoo.com (8.13.8/8.13.8/y.out) with ESMTP id l076bLe8007164; Sat, 6 Jan 2007 22:37:22 -0800 (PST) Date: Sat, 06 Jan 2007 10:17:08 +0900 Message-ID: From: gnn@freebsd.org To: Eugene Grosbein In-Reply-To: <20070105174350.GA21615@svzserv.kemerovo.su> References: <20070105174350.GA21615@svzserv.kemerovo.su> User-Agent: Wanderlust/2.14.0 (Africa) SEMI/1.14.6 (Maruoka) FLIM/1.14.8 (=?ISO-8859-4?Q?Shij=F2?=) APEL/10.6 Emacs/22.0.90 (i386-apple-darwin8.8.1) MULE/5.0 (SAKAKI) MIME-Version: 1.0 (generated by SEMI 1.14.6 - "Maruoka") Content-Type: text/plain; charset=US-ASCII Cc: stable@freebsd.org, performance@freebsd.org Subject: Re: benchmark X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sun, 07 Jan 2007 06:48:05 -0000 You should try ports/net/netpipe which has the nice side effect of shoving different message sizes across, and tends to show lots of interesting performance issues. Best, George From owner-freebsd-performance@FreeBSD.ORG Fri Jan 12 09:27:51 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id D238216A40F for ; Fri, 12 Jan 2007 09:27:51 +0000 (UTC) (envelope-from patpro@patpro.net) Received: from smtp.univ-lyon2.fr (smtp.univ-lyon2.fr [159.84.143.102]) by mx1.freebsd.org (Postfix) with ESMTP id 92F1D13C442 for ; Fri, 12 Jan 2007 09:27:51 +0000 (UTC) (envelope-from patpro@patpro.net) Received: from localhost (localhost [127.0.0.1]) by smtp.univ-lyon2.fr (Postfix) with ESMTP id AE3121861833 for ; Fri, 12 Jan 2007 10:00:50 +0100 (CET) X-Virus-Scanned: amavisd-new at univ-lyon2.fr Received: from smtp.univ-lyon2.fr ([127.0.0.1]) by localhost (smtp.univ-lyon2.fr [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id zaomcp140n4L for ; Fri, 12 Jan 2007 10:00:48 +0100 (CET) Received: from [159.84.148.59] (patpro.univ-lyon2.fr [159.84.148.59]) by smtp.univ-lyon2.fr (Postfix) with ESMTP id CD7CC1861810 for ; Fri, 12 Jan 2007 10:00:48 +0100 (CET) Mime-Version: 1.0 (Apple Message framework v752.2) Content-Transfer-Encoding: 7bit Message-Id: Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed To: freebsd-performance@freebsd.org From: Patrick Proniewski Date: Fri, 12 Jan 2007 10:00:47 +0100 X-Mailer: Apple Mail (2.752.2) Subject: network perf : em driver ? X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 12 Jan 2007 09:27:51 -0000 Hello, I'm the happy owner of a Tyan Tiger i7520SD motherboard, sporting two ethernet ports using a Intel 82571EB GbE controller and running FreeBSD 6.2 RC1. FreeBSD uses the em driver for this ports: dmesg excerpt: em0: port 0x2000-0x201f mem 0xd8020000-0xd803ffff,0xd8000000-0xd801ffff irq 16 at device 0.0 on pci3 ifconfig output: em0: flags=8843 mtu 1500 options=b inet 192.168.0.1 netmask 0xffffff00 broadcast 192.168.0.255 ether 00:e0:81:42:e3:aa media: Ethernet autoselect (1000baseTX ) status: active The box uses its 100baseT fxp driven eth port for internet access, and shares this access with a Mac OS X plugged on the em0 interface. I've set up pf with firewall and nat rules to act as a gateway. Both machines are in "Ethernet autoselect (1000baseTX )" and linked with a direct ethernet wire. File transfert through em0 looks quite slow : at best 17 MB/s (scp gives better results than http, tested with a 120MB file) File copy, from one HDD to another or from one HDD to /dev/null, ranges from 30 to 48 MB/s on the FreeBSD side (SATA HD). I would like to know if there is a good method to find the bottleneck and to get rid of it. How comes my GbE is so slow ? thanks, patpro From owner-freebsd-performance@FreeBSD.ORG Fri Jan 12 09:52:40 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id 7193516A407 for ; Fri, 12 Jan 2007 09:52:40 +0000 (UTC) (envelope-from dom@helenmarks.co.uk) Received: from mailhost.graphdata.co.uk (mailhost.graphdata.co.uk [195.12.22.194]) by mx1.freebsd.org (Postfix) with ESMTP id 2EB4913C455 for ; Fri, 12 Jan 2007 09:52:40 +0000 (UTC) (envelope-from dom@helenmarks.co.uk) Received: from localhost (localhost [127.0.0.1]) by mailhost.graphdata.co.uk (Postfix) with ESMTP id 7856C114020; Fri, 12 Jan 2007 09:35:15 +0000 (GMT) X-Virus-Scanned: amavisd-new at graphdata.co.uk Received: from mailhost.graphdata.co.uk ([127.0.0.1]) by localhost (mailhost.graphdata.co.uk [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 1SOjhPi7jhA3; Fri, 12 Jan 2007 09:35:12 +0000 (GMT) Received: from gdc083.internal.graphdata.co.uk (gdc083.internal.graphdata.co.uk [192.168.0.86]) by mailhost.graphdata.co.uk (Postfix) with SMTP id D9B3F11401E; Fri, 12 Jan 2007 09:35:12 +0000 (GMT) Date: Fri, 12 Jan 2007 09:35:12 +0000 From: Dominic Marks To: Patrick Proniewski Message-Id: <20070112093512.83dd49fd.dom@helenmarks.co.uk> In-Reply-To: References: X-Mailer: Sylpheed 2.3.0 (GTK+ 2.10.6; i386-portbld-freebsd6.2) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Cc: freebsd-performance@freebsd.org Subject: Re: network perf : em driver ? X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 12 Jan 2007 09:52:40 -0000 On Fri, 12 Jan 2007 10:00:47 +0100 Patrick Proniewski wrote: > Hello, > > I'm the happy owner of a Tyan Tiger i7520SD motherboard, sporting two > ethernet ports using a Intel 82571EB GbE controller and running > FreeBSD 6.2 RC1. FreeBSD uses the em driver for this ports: > > dmesg excerpt: > em0: port > 0x2000-0x201f mem 0xd8020000-0xd803ffff,0xd8000000-0xd801ffff irq 16 > at device 0.0 on pci3 > > ifconfig output: > em0: flags=8843 mtu 1500 > options=b > inet 192.168.0.1 netmask 0xffffff00 broadcast 192.168.0.255 > ether 00:e0:81:42:e3:aa > media: Ethernet autoselect (1000baseTX ) > status: active > > The box uses its 100baseT fxp driven eth port for internet access, > and shares this access with a Mac OS X plugged on the em0 interface. > I've set up pf with firewall and nat rules to act as a gateway. Both > machines are in "Ethernet autoselect (1000baseTX )" and > linked with a direct ethernet wire. > > File transfert through em0 looks quite slow : at best 17 MB/s (scp > gives better results than http, tested with a 120MB file) > File copy, from one HDD to another or from one HDD to /dev/null, > ranges from 30 to 48 MB/s on the FreeBSD side (SATA HD). Try testing FTP. SCP will be doing encryption which will slow the process down. I can't say which HTTP would be slower than that. I have some Dell's with similar specification which will happily FTP back and forth at around 40MB/s which is reasonable. > I would like to know if there is a good method to find the bottleneck > and to get rid of it. How comes my GbE is so slow ? This is a dedicated network right? No other traffic? Dom From owner-freebsd-performance@FreeBSD.ORG Fri Jan 12 10:32:53 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id A1C4C16A412 for ; Fri, 12 Jan 2007 10:32:53 +0000 (UTC) (envelope-from patpro@patpro.net) Received: from smtp.univ-lyon2.fr (smtp.univ-lyon2.fr [159.84.143.102]) by mx1.freebsd.org (Postfix) with ESMTP id 6284F13C448 for ; Fri, 12 Jan 2007 10:32:53 +0000 (UTC) (envelope-from patpro@patpro.net) Received: from localhost (localhost [127.0.0.1]) by smtp.univ-lyon2.fr (Postfix) with ESMTP id 6167918680C2; Fri, 12 Jan 2007 11:32:52 +0100 (CET) X-Virus-Scanned: amavisd-new at univ-lyon2.fr Received: from smtp.univ-lyon2.fr ([127.0.0.1]) by localhost (smtp.univ-lyon2.fr [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id TGX+n9+JtKN3; Fri, 12 Jan 2007 11:32:51 +0100 (CET) Received: from [159.84.148.59] (patpro.univ-lyon2.fr [159.84.148.59]) by smtp.univ-lyon2.fr (Postfix) with ESMTP id 44C4618680B9; Fri, 12 Jan 2007 11:32:51 +0100 (CET) In-Reply-To: <20070112093512.83dd49fd.dom@helenmarks.co.uk> References: <20070112093512.83dd49fd.dom@helenmarks.co.uk> Mime-Version: 1.0 (Apple Message framework v752.2) Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: Content-Transfer-Encoding: 7bit From: Patrick Proniewski Date: Fri, 12 Jan 2007 11:32:50 +0100 To: Dominic Marks X-Mailer: Apple Mail (2.752.2) Cc: freebsd-performance@freebsd.org Subject: Re: network perf : em driver ? X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 12 Jan 2007 10:32:53 -0000 Hi Dominic, On 12 janv. 07, at 10:35, Dominic Marks wrote: >> File transfert through em0 looks quite slow : at best 17 MB/s (scp >> gives better results than http, tested with a 120MB file) >> File copy, from one HDD to another or from one HDD to /dev/null, >> ranges from 30 to 48 MB/s on the FreeBSD side (SATA HD). > > Try testing FTP. SCP will be doing encryption which will slow > the process down. I can't say which HTTP would be slower than > that. I'm not sure encryption slows it down, the box has 4 CPU (2 dual core Xeon LV 1.66 GHz), and as I've said, non-encrypted HTTP was slower :/ > I have some Dell's with similar specification which will > happily FTP back and forth at around 40MB/s which is reasonable. I'll give FTP a try, but I would like the network to be fast for every protocols. I'm planning to share data using NFS, WebDAV, or SMB (and scp occasionally), but I've still to choose and configure appropriate servers. >> I would like to know if there is a good method to find the bottleneck >> and to get rid of it. How comes my GbE is so slow ? > > This is a dedicated network right? No other traffic? the Mac OS X box access the Internet via the FreeBSD gateway (mac<-- >em0(freebsd)fxp0<-->internet), so it's not absolutely dedicated to my network benchmarks. But internet related traffic was as low as possible during my tests (few imap sessions, one irc connexion). patpro From owner-freebsd-performance@FreeBSD.ORG Fri Jan 12 10:45:43 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id C01A916A403 for ; Fri, 12 Jan 2007 10:45:43 +0000 (UTC) (envelope-from arne_woerner@yahoo.com) Received: from web30309.mail.mud.yahoo.com (web30309.mail.mud.yahoo.com [209.191.69.71]) by mx1.freebsd.org (Postfix) with SMTP id 71B4213C467 for ; Fri, 12 Jan 2007 10:45:43 +0000 (UTC) (envelope-from arne_woerner@yahoo.com) Received: (qmail 106 invoked by uid 60001); 12 Jan 2007 10:45:42 -0000 DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.com; h=X-YMail-OSG:Received:Date:From:Subject:To:Cc:In-Reply-To:MIME-Version:Content-Type:Content-Transfer-Encoding:Message-ID; b=f9sxZP76qyR23x4teDWlQgS3xmWeQ99t/qZHd8RWtlKqOZ8qtONozs0qYjwaEt0HtzYqkXBuHq0tdLy0bKpScLhIGDu7KV9XAIH7xk/WbNtbdjJKdYiV2hWxjCpxNCf2u2igyLBvoJvvLFFtWDrAXteZbLHmu8bdYdLSx3KCrm0=; X-YMail-OSG: lm5XCxEVM1lavbR7_f0hrzmn8WmWJXRwITmh3oJh.K_yIh_9waA3FedRcC0DbzpHkIj6jilHb8qpFco__tMO5djlU.zuAHUwVzS7qvt53c0MadedgFaGBi8AhS58iyqppBIURzZdY5I- Received: from [85.212.6.114] by web30309.mail.mud.yahoo.com via HTTP; Fri, 12 Jan 2007 02:45:42 PST Date: Fri, 12 Jan 2007 02:45:42 -0800 (PST) From: "R. B. Riddick" To: Patrick Proniewski In-Reply-To: MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit Message-ID: <379658.99357.qm@web30309.mail.mud.yahoo.com> Cc: freebsd-performance@freebsd.org Subject: Re: network perf : em driver ? X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 12 Jan 2007 10:45:43 -0000 --- Patrick Proniewski wrote: > I'll give FTP a try, but I would like the network to be fast for > every protocols. I'm planning to share data using NFS, WebDAV, or SMB > (and scp occasionally), but I've still to choose and configure > appropriate servers. > We had that problem before: Some HTTP server implementations just dont bring it... :-) thttpd is quite efficient, I have heard... You can try 1. src/tools/tools/netrate/netblast 2. increase MTU (ifconfig em0 mtu 65536 or so; never tried that myself) 3. ports/benchmarks/tcpblast 4. build something with nc: server: nc -l 1234 > /dev/null client: dd if=/dev/zero bs=1m | nc serverIP 1234 which will eliminate disk latency... -Arne ____________________________________________________________________________________ 8:00? 8:25? 8:40? Find a flick in no time with the Yahoo! Search movie showtime shortcut. http://tools.search.yahoo.com/shortcuts/#news From owner-freebsd-performance@FreeBSD.ORG Fri Jan 12 12:53:58 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id 53C0816A40F for ; Fri, 12 Jan 2007 12:53:58 +0000 (UTC) (envelope-from patpro@patpro.net) Received: from smtp.univ-lyon2.fr (smtp.univ-lyon2.fr [159.84.143.102]) by mx1.freebsd.org (Postfix) with ESMTP id 1427213C458 for ; Fri, 12 Jan 2007 12:53:58 +0000 (UTC) (envelope-from patpro@patpro.net) Received: from localhost (localhost [127.0.0.1]) by smtp.univ-lyon2.fr (Postfix) with ESMTP id 1C6461871025; Fri, 12 Jan 2007 13:53:57 +0100 (CET) X-Virus-Scanned: amavisd-new at univ-lyon2.fr Received: from smtp.univ-lyon2.fr ([127.0.0.1]) by localhost (smtp.univ-lyon2.fr [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id z6CtoVbeEibv; Fri, 12 Jan 2007 13:53:55 +0100 (CET) Received: from [159.84.148.59] (patpro.univ-lyon2.fr [159.84.148.59]) by smtp.univ-lyon2.fr (Postfix) with ESMTP id D51971871013; Fri, 12 Jan 2007 13:53:55 +0100 (CET) In-Reply-To: <379658.99357.qm@web30309.mail.mud.yahoo.com> References: <379658.99357.qm@web30309.mail.mud.yahoo.com> Mime-Version: 1.0 (Apple Message framework v752.2) Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: <9E3FD55C-FA00-4909-8A52-AA9F46F9BE65@patpro.net> Content-Transfer-Encoding: 7bit From: Patrick Proniewski Date: Fri, 12 Jan 2007 13:53:55 +0100 To: R.B.Riddick X-Mailer: Apple Mail (2.752.2) Cc: freebsd-performance@freebsd.org Subject: Re: network perf : em driver ? X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 12 Jan 2007 12:53:58 -0000 On 12 janv. 07, at 11:45, R. B. Riddick wrote: > --- Patrick Proniewski wrote: >> I'll give FTP a try, but I would like the network to be fast for >> every protocols. I'm planning to share data using NFS, WebDAV, or SMB >> (and scp occasionally), but I've still to choose and configure >> appropriate servers. >> > We had that problem before: Some HTTP server implementations just > dont bring > it... :-) thttpd is quite efficient, I have heard... apache (2.2) is not supposed to be so inefficient :/ But you are right, I might need to try a lighter server. > You can try > 1. src/tools/tools/netrate/netblast > 2. increase MTU (ifconfig em0 mtu 65536 or so; never tried that > myself) > 3. ports/benchmarks/tcpblast > 4. build something with nc: > server: nc -l 1234 > /dev/null > client: dd if=/dev/zero bs=1m | nc serverIP 1234 > which will eliminate disk latency... I'll try this ASAP. thank you for the tips. patpro From owner-freebsd-performance@FreeBSD.ORG Fri Jan 12 13:44:13 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id 6C58016A407 for ; Fri, 12 Jan 2007 13:44:13 +0000 (UTC) (envelope-from Rich@WhiteOakLabs.com) Received: from whiteoaklabs.com (mail.whiteoaklabs.com [69.55.226.191]) by mx1.freebsd.org (Postfix) with ESMTP id 370AB13C448 for ; Fri, 12 Jan 2007 13:44:11 +0000 (UTC) (envelope-from Rich@WhiteOakLabs.com) Received: from xa.home.org (cpe-24-174-80-60.houston.res.rr.com [24.174.80.60]) by whiteoaklabs.com (8.12.11/8.12.11) with ESMTP id l0CDSXqs031841 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=OK) for ; Fri, 12 Jan 2007 05:28:33 -0800 Received: from [172.16.1.6] (a.home.org [172.16.1.6]) by xa.home.org (8.13.6/8.13.4) with ESMTP id l0CDUhZ8056308 for ; Fri, 12 Jan 2007 07:30:43 -0600 (CST) (envelope-from Rich@WhiteOakLabs.com) Message-ID: <45A78CFC.1080607@WhiteOakLabs.com> Date: Fri, 12 Jan 2007 07:28:28 -0600 From: Rich Murphey User-Agent: Thunderbird 1.5.0.9 (Windows/20061207) MIME-Version: 1.0 To: freebsd-performance@freebsd.org References: <379658.99357.qm@web30309.mail.mud.yahoo.com> In-Reply-To: <379658.99357.qm@web30309.mail.mud.yahoo.com> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-DCC-sonic.net-Metrics: whiteoaklabs.com 1117; Body=1 Fuz1=1 Fuz2=1 Subject: Re: network perf : em driver ? X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 12 Jan 2007 13:44:13 -0000 R. B. Riddick wrote: > --- Patrick Proniewski wrote: > >> I'll give FTP a try, but I would like the network to be fast for >> every protocols. I'm planning to share data using NFS, WebDAV, or SMB >> (and scp occasionally), but I've still to choose and configure >> appropriate servers. >> >> > We had that problem before: Some HTTP server implementations just dont bring > it... :-) thttpd is quite efficient, I have heard... > > You can try > 1. src/tools/tools/netrate/netblast > 2. increase MTU (ifconfig em0 mtu 65536 or so; never tried that myself) > 3. ports/benchmarks/tcpblast > 4. build something with nc: > server: nc -l 1234 > /dev/null > client: dd if=/dev/zero bs=1m | nc serverIP 1234 > which will eliminate disk latency... > > -Arne > lighttpd has good support for kqueue and sendfile, so depending on whether the measurement will include lots of http connections or static files, it may be worth considering. It's feature set and code size are somewhat larger than thttpd I believe. Rich From owner-freebsd-performance@FreeBSD.ORG Fri Jan 12 15:43:13 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id 2539216A412 for ; Fri, 12 Jan 2007 15:43:13 +0000 (UTC) (envelope-from gofp-freebsd-performance@m.gmane.org) Received: from ciao.gmane.org (main.gmane.org [80.91.229.2]) by mx1.freebsd.org (Postfix) with ESMTP id D8A8613C455 for ; Fri, 12 Jan 2007 15:43:12 +0000 (UTC) (envelope-from gofp-freebsd-performance@m.gmane.org) Received: from list by ciao.gmane.org with local (Exim 4.43) id 1H5OYZ-0005At-0v for freebsd-performance@freebsd.org; Fri, 12 Jan 2007 16:43:03 +0100 Received: from lara.cc.fer.hr ([161.53.72.113]) by main.gmane.org with esmtp (Gmexim 0.1 (Debian)) id 1AlnuQ-0007hv-00 for ; Fri, 12 Jan 2007 16:43:03 +0100 Received: from ivoras by lara.cc.fer.hr with local (Gmexim 0.1 (Debian)) id 1AlnuQ-0007hv-00 for ; Fri, 12 Jan 2007 16:43:03 +0100 X-Injected-Via-Gmane: http://gmane.org/ To: freebsd-performance@freebsd.org From: Ivan Voras Date: Fri, 12 Jan 2007 16:42:48 +0100 Lines: 24 Message-ID: References: <379658.99357.qm@web30309.mail.mud.yahoo.com> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit X-Complaints-To: usenet@sea.gmane.org X-Gmane-NNTP-Posting-Host: lara.cc.fer.hr User-Agent: Thunderbird 1.5.0.4 (X11/20060625) In-Reply-To: <379658.99357.qm@web30309.mail.mud.yahoo.com> Sender: news Subject: Re: network perf : em driver ? X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 12 Jan 2007 15:43:13 -0000 R. B. Riddick wrote: > We had that problem before: Some HTTP server implementations just dont bring > it... :-) thttpd is quite efficient, I have heard... This is a red herring. The OP reports he transfers a single file - http server performance cannot even approach to influence the performance in his case (and he's using apache!). There's absolutely no reason to replace apache here. > You can try > 2. increase MTU (ifconfig em0 mtu 65536 or so; never tried that myself) Don't do that. His ifconfig output lists his card doesn't support jumbo frames, and most ethernet card's will wedge if you use MTU > 1500, let alone a nonstandard one as 65536. Even in 1999, standard 1500 byte frames could yield 400 Mbit/s (http://sd.wareonearth.com/~phil/jumbo.html) > 3. ports/benchmarks/tcpblast I've found iperf to be more useful. From owner-freebsd-performance@FreeBSD.ORG Fri Jan 12 16:07:22 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id 5F74C16A415 for ; Fri, 12 Jan 2007 16:07:22 +0000 (UTC) (envelope-from arne_woerner@yahoo.com) Received: from web30312.mail.mud.yahoo.com (web30312.mail.mud.yahoo.com [209.191.69.74]) by mx1.freebsd.org (Postfix) with SMTP id 2908713C442 for ; Fri, 12 Jan 2007 16:07:22 +0000 (UTC) (envelope-from arne_woerner@yahoo.com) Received: (qmail 40992 invoked by uid 60001); 12 Jan 2007 16:07:21 -0000 DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.com; h=X-YMail-OSG:Received:Date:From:Subject:To:In-Reply-To:MIME-Version:Content-Type:Content-Transfer-Encoding:Message-ID; b=xM25fv3Hfu8s/Vf1pFZki9VokLyg6Yb8ltsicARHgORGi5Kxrp2R+zzrBiGQmRK0IYQ3gxPF8VC3z6flOW972KGENDY2C+vHENOXOvAhaO1/EJ9XAgCNvEcXIxZQDwVoB+/T7qR737app03Kr0Z5zaXogrnptkwtPOL/yTd9fS8=; X-YMail-OSG: XGJbyi8VM1lUbm86BkfOdeln0DJvscxxZ34BFv6ZtwoW9HlDxTsEZ3W4n0NhibTLhnEQcVdDtx1X7ucDMtP6jguBT3JOmxb5wrKEfH_utEUxZZzxc5cmpFre4nhcE.GB2_G69rCW.wy2mSs- Received: from [85.212.6.114] by web30312.mail.mud.yahoo.com via HTTP; Fri, 12 Jan 2007 08:07:21 PST Date: Fri, 12 Jan 2007 08:07:21 -0800 (PST) From: "R. B. Riddick" To: freebsd-performance@freebsd.org In-Reply-To: MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit Message-ID: <465799.40884.qm@web30312.mail.mud.yahoo.com> Subject: Re: network perf : em driver ? X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 12 Jan 2007 16:07:22 -0000 --- Ivan Voras wrote: > R. B. Riddick wrote: > > We had that problem before: Some HTTP server implementations just dont > > bring it... :-) thttpd is quite efficient, I have heard... > > This is a red herring. The OP reports he transfers a single file - http > server performance cannot even approach to influence the performance in > his case (and he's using apache!). There's absolutely no reason to > replace apache here. > Soso... Those are words and no herrings (especially no red herrings); their colour should be black in most cases... As the "OP" (what is that exactly? again an animal?) mentioned: Apache performs worse than scp. My memory tells me similar things... Remember: Apache is optimized for LINUX not necessarily for FreeBSD... I am pretty sure, that I could write an http server, that does not use more than 1% of a Gigabit/connection... So there might be a very good reason to replace Apache (2.2 or what it was) by something, that is smarter on FreeBSD (like scp with http capability but without encryption)... So we want to conclude, that I do not apply tricks or feints, and that I do not have any reason to do so... I just reflect here, what I have learned... pro bono publico hopefully... > > You can try > > > 2. increase MTU (ifconfig em0 mtu 65536 or so; never tried that myself) > > Don't do that. His ifconfig output lists his card doesn't support jumbo > frames, and most ethernet card's will wedge if you use MTU > 1500, let > alone a nonstandard one as 65536. > > Even in 1999, standard 1500 byte frames could yield 400 Mbit/s > (http://sd.wareonearth.com/~phil/jumbo.html) > Oh OK... I wasnt aware that there are Gbit NICs, that cant do big mtu-s... > > 3. ports/benchmarks/tcpblast > > I've found iperf to be more useful. > Soso... Was it slower? Or what? -Arne ____________________________________________________________________________________ Food fight? Enjoy some healthy debate in the Yahoo! Answers Food & Drink Q&A. http://answers.yahoo.com/dir/?link=list&sid=396545367 From owner-freebsd-performance@FreeBSD.ORG Fri Jan 12 17:56:33 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id 7559316A407 for ; Fri, 12 Jan 2007 17:56:33 +0000 (UTC) (envelope-from lists@wm-access.no) Received: from lakepoint.domeneshop.no (smtp01.domeneshop.no [194.63.248.15]) by mx1.freebsd.org (Postfix) with ESMTP id 0FFC413C441 for ; Fri, 12 Jan 2007 17:56:32 +0000 (UTC) (envelope-from lists@wm-access.no) Received: from [192.168.0.100] (225.0.33.65.cfl.res.rr.com [65.33.0.225]) (authenticated bits=0) by lakepoint.domeneshop.no (8.13.8/8.13.8) with ESMTP id l0CHajln003302 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NOT); Fri, 12 Jan 2007 18:36:47 +0100 Message-ID: <45A7C72C.6080404@wm-access.no> Date: Fri, 12 Jan 2007 12:36:44 -0500 From: =?ISO-8859-1?Q?Sten_Daniel_S=F8rsdal?= User-Agent: Thunderbird 1.5.0.9 (Windows/20061207) MIME-Version: 1.0 To: Patrick Proniewski References: <379658.99357.qm@web30309.mail.mud.yahoo.com> <9E3FD55C-FA00-4909-8A52-AA9F46F9BE65@patpro.net> In-Reply-To: <9E3FD55C-FA00-4909-8A52-AA9F46F9BE65@patpro.net> Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Cc: freebsd-performance@freebsd.org, "R.B.Riddick" Subject: Re: network perf : em driver ? X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 12 Jan 2007 17:56:33 -0000 Patrick Proniewski wrote: > On 12 janv. 07, at 11:45, R. B. Riddick wrote: >=20 >> --- Patrick Proniewski wrote: >>> I'll give FTP a try, but I would like the network to be fast for >>> every protocols. I'm planning to share data using NFS, WebDAV, or SMB= >>> (and scp occasionally), but I've still to choose and configure >>> appropriate servers. >>> >> We had that problem before: Some HTTP server implementations just dont= >> bring >> it... :-) thttpd is quite efficient, I have heard... >=20 > apache (2.2) is not supposed to be so inefficient :/ But you are right,= > I might need to try a lighter server. >=20 >> You can try >> 1. src/tools/tools/netrate/netblast >> 2. increase MTU (ifconfig em0 mtu 65536 or so; never tried that myself= ) >> 3. ports/benchmarks/tcpblast >> 4. build something with nc: >> server: nc -l 1234 > /dev/null >> client: dd if=3D/dev/zero bs=3D1m | nc serverIP 1234 >> which will eliminate disk latency... >=20 >=20 > I'll try this ASAP. > thank you for the tips. >=20 Maybe i misunderstood something but i think 6.2 has some changes done to how it identifies local networks and how that affects inflight. Try disabling tcp inflight for these tests. Also SACK seems to affect the local network performance. Also try: sysctl net.inet.tcp.delayedack=3D0 --=20 Sten Daniel S=F8rsdal From owner-freebsd-performance@FreeBSD.ORG Fri Jan 12 18:22:34 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id B8E7316A416 for ; Fri, 12 Jan 2007 18:22:34 +0000 (UTC) (envelope-from amesbury@umn.edu) Received: from mta-a2.tc.umn.edu (mta-a2.tc.umn.edu [134.84.119.206]) by mx1.freebsd.org (Postfix) with ESMTP id 711E213C469 for ; Fri, 12 Jan 2007 18:22:34 +0000 (UTC) (envelope-from amesbury@umn.edu) Received: from [160.94.247.212] (paulaner.oitsec.umn.edu [160.94.247.212]) by mta-a2.tc.umn.edu (UMN smtpd) with ESMTP for ; Fri, 12 Jan 2007 12:07:33 -0600 (CST) X-Umn-Remote-Mta: [N] paulaner.oitsec.umn.edu [160.94.247.212] #+LO+TS+AU+HN Message-ID: <45A7CE64.4000800@umn.edu> Date: Fri, 12 Jan 2007 12:07:32 -0600 From: Alan Amesbury User-Agent: Thunderbird 1.5.0.9 (X11/20061222) MIME-Version: 1.0 To: freebsd-performance@freebsd.org References: <20070112120037.1797516A5DB@hub.freebsd.org> In-Reply-To: <20070112120037.1797516A5DB@hub.freebsd.org> X-Enigmail-Version: 0.94.0.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Subject: Re: network perf : em driver ? X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 12 Jan 2007 18:22:34 -0000 "R. B. Riddick" wrote: > --- Patrick Proniewski wrote: >> I'll give FTP a try, but I would like the network to be fast for >> every protocols. I'm planning to share data using NFS, WebDAV, or SMB >> (and scp occasionally), but I've still to choose and configure >> appropriate servers. >> > We had that problem before: Some HTTP server implementations just dont bring > it... :-) thttpd is quite efficient, I have heard... > > You can try > 1. src/tools/tools/netrate/netblast > 2. increase MTU (ifconfig em0 mtu 65536 or so; never tried that myself) I don't think you want to do this, as *all* devices on the same network segment (layer 2) will have to use the same MTU for it to work safely and reliably. Besides, I think em(4)'s maximum MTU is 9216, so I don't think you *can* set an MTU that high. Again, if you're changing MTU, make sure that everything else on that same segment changes MTU as well. Besides, even with an MTU of 1500, a gigabit network should be able to beat 17MB/sec. > 3. ports/benchmarks/tcpblast > 4. build something with nc: > server: nc -l 1234 > /dev/null > client: dd if=/dev/zero bs=1m | nc serverIP 1234 > which will eliminate disk latency... I initially thought scp's encryption and compression overhead were possible sources of throughput problems, but Patrick (in his initial posting) said that scp is *faster* than HTTP. Since SMP is apparently involved, I'm wondering if this is related to the various em(4) problems noted earlier in a number of threads on -stable and -hackers. I'd suggest checking those for clues, particularly if SMP actually is being used. Polling may also help... or may not; I think it's dependent on the load and task at hand. -- Alan Amesbury OIT Security and Assurance University of Minnesota From owner-freebsd-performance@FreeBSD.ORG Fri Jan 12 18:30:10 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id 0C49B16A403 for ; Fri, 12 Jan 2007 18:30:10 +0000 (UTC) (envelope-from arne_woerner@yahoo.com) Received: from web30306.mail.mud.yahoo.com (web30306.mail.mud.yahoo.com [209.191.69.68]) by mx1.freebsd.org (Postfix) with SMTP id BB9E413C45D for ; Fri, 12 Jan 2007 18:30:07 +0000 (UTC) (envelope-from arne_woerner@yahoo.com) Received: (qmail 41609 invoked by uid 60001); 12 Jan 2007 18:30:07 -0000 DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.com; h=Message-ID:X-YMail-OSG:Received:Date:From:Subject:To:Cc:In-Reply-To:MIME-Version:Content-Type:Content-Transfer-Encoding; b=rz6dSswoaMEWizDunPfWxHD+IMsXVf47amxrMmjcmRGEY59hM4skBekV9/QGDDeFgsivcUyYCC9KOTHflsQij+t8hooXi5aHeMtllnLh29RhqJHfCOYenTXN+Rx8mqgtue3wRxSTJWi3XEXFcLFLVc2xyeQNzdPW6QydenGwwMs= ; Message-ID: <20070112183007.41607.qmail@web30306.mail.mud.yahoo.com> X-YMail-OSG: nUBoBRAVM1mVkLtzI2sM9_xALM1atPK0aSDZnt3dx4PnTMytKr3uMr9QHXR9Arli4nmOpk9WDxe7WCou3zaqXNWhqBFtXQoVba9w4JfZEr4E4SMV.VzaTWtEkujPNzH9UzSDgvxMwTvMgA6GQDIvK0s_qMkntQZPA4sjMIPYLy1cgGdEgV6eTiOwVIWc Received: from [85.212.6.114] by web30306.mail.mud.yahoo.com via HTTP; Fri, 12 Jan 2007 10:30:07 PST Date: Fri, 12 Jan 2007 10:30:07 -0800 (PST) From: "R. B. Riddick" To: Chuck Swiger In-Reply-To: MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit Cc: freebsd-performance@freebsd.org Subject: Re: network perf : em driver ? X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 12 Jan 2007 18:30:10 -0000 --- Chuck Swiger wrote: > On Jan 12, 2007, at 8:07 AM, R. B. Riddick wrote: > > As the "OP" (what is that exactly? again an animal?) mentioned: > > Apache performs > > worse than scp. > > Quick testing suggests that an Apache child process accumulates a > similar amount of CPU time transferring large files as scp when using > an SSL connection; if you access Apache via HTTP rather than HTTPS, > it uses much less CPU than scp does. > Hmm... CPU usage isnt all that counts... I can write an http server, that uses 0.1% CPU time and produces 800bit/sec traffic... :-) > > My memory tells me similar things... Remember: Apache is > > optimized for LINUX not necessarily for FreeBSD... > > Apache's been optimized to run quite well on a lot of platforms, > although it is somewhat heavy-weight compared with a webserver > oriented towards serving static files only. > I just cant find the thread about apache/thttpd (it was in freebsd-performance@, I think), and I dont have up-to-date hardware, and I am too lazy to compare Apache on FreeBSD to the theoretical maximum, some piece of software can reach on FreeBSD... Since these systems are nowadays so complex and difficult to compare, I would recommend benchmarks, that r tailored on and for the special box... Ideally one would use different boxes and test each with different OSes and different application-implementations, so that u can do an informed decision based on empirical data in the end (of course I know, that limited time and funding causes some pressure)... -Arne --- "Denny Crane, Denny, Denny, Denny ... Crane" (from "Boston Legal") "Get me Homer Landskirty" (from "Scary Movie 4") ____________________________________________________________________________________ No need to miss a message. Get email on-the-go with Yahoo! Mail for Mobile. Get started. http://mobile.yahoo.com/mail From owner-freebsd-performance@FreeBSD.ORG Fri Jan 12 18:41:14 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id 83D3516A40F for ; Fri, 12 Jan 2007 18:41:14 +0000 (UTC) (envelope-from cswiger@mac.com) Received: from mail-out4.apple.com (mail-out4.apple.com [17.254.13.23]) by mx1.freebsd.org (Postfix) with ESMTP id 6BD4913C455 for ; Fri, 12 Jan 2007 18:41:12 +0000 (UTC) (envelope-from cswiger@mac.com) Received: from relay6.apple.com (relay6.apple.com [17.128.113.36]) by mail-out4.apple.com (8.13.8/8.13.8) with ESMTP id l0CHoiWp027763; Fri, 12 Jan 2007 09:50:44 -0800 (PST) Received: from relay6.apple.com (unknown [127.0.0.1]) by relay6.apple.com (Symantec Mail Security) with ESMTP id 2532A1009A; Fri, 12 Jan 2007 09:50:44 -0800 (PST) X-AuditID: 11807124-a1ce9bb000006d75-7d-45a7ca744103 Received: from [17.214.13.96] (unknown [17.214.13.96]) (using TLSv1 with cipher AES128-SHA (128/128 bits)) (No client certificate requested) by relay6.apple.com (Apple SCV relay) with ESMTP id 10C0310084; Fri, 12 Jan 2007 09:50:44 -0800 (PST) In-Reply-To: <465799.40884.qm@web30312.mail.mud.yahoo.com> References: <465799.40884.qm@web30312.mail.mud.yahoo.com> Mime-Version: 1.0 (Apple Message framework v752.2) Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: Content-Transfer-Encoding: 7bit From: Chuck Swiger Date: Fri, 12 Jan 2007 09:50:41 -0800 To: "R. B. Riddick" X-Mailer: Apple Mail (2.752.2) X-Brightmail-Tracker: AAAAAA== Cc: freebsd-performance@freebsd.org Subject: Re: network perf : em driver ? X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 12 Jan 2007 18:41:14 -0000 On Jan 12, 2007, at 8:07 AM, R. B. Riddick wrote: > As the "OP" (what is that exactly? again an animal?) mentioned: > Apache performs > worse than scp. Quick testing suggests that an Apache child process accumulates a similar amount of CPU time transferring large files as scp when using an SSL connection; if you access Apache via HTTP rather than HTTPS, it uses much less CPU than scp does. > My memory tells me similar things... Remember: Apache is > optimized for LINUX not necessarily for FreeBSD... Apache's been optimized to run quite well on a lot of platforms, although it is somewhat heavy-weight compared with a webserver oriented towards serving static files only. -- -Chuck From owner-freebsd-performance@FreeBSD.ORG Fri Jan 12 23:37:14 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id D00BF16A403 for ; Fri, 12 Jan 2007 23:37:14 +0000 (UTC) (envelope-from patpro@patpro.net) Received: from postfix1-g20.free.fr (postfix1-g20.free.fr [212.27.60.42]) by mx1.freebsd.org (Postfix) with ESMTP id 460AA13C455 for ; Fri, 12 Jan 2007 23:37:14 +0000 (UTC) (envelope-from patpro@patpro.net) Received: from smtp1-g19.free.fr (smtp1-g19.free.fr [212.27.42.27]) by postfix1-g20.free.fr (Postfix) with ESMTP id 46E507AA266 for ; Sat, 13 Jan 2007 00:08:49 +0100 (CET) Received: from [192.168.0.2] (boleskine.patpro.net [82.235.12.223]) by smtp1-g19.free.fr (Postfix) with ESMTP id 1AEAD9B650; Sat, 13 Jan 2007 00:08:47 +0100 (CET) In-Reply-To: References: Mime-Version: 1.0 (Apple Message framework v752.2) Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: <7CA8AE1A-3925-404C-9F69-32AC4FFBB379@patpro.net> Content-Transfer-Encoding: 7bit From: Patrick Proniewski Date: Sat, 13 Jan 2007 00:08:46 +0100 To: Lucas Holt X-Mailer: Apple Mail (2.752.2) Cc: freebsd-performance@freebsd.org Subject: Re: network perf : em driver ? X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 12 Jan 2007 23:37:14 -0000 On 12 janv. 2007, at 23:01, Lucas Holt wrote: > The first two things I would try beyond the sysctls mentioned would > be to try another network cable, ok, added to the todo list ;) > and view your firewall configuration on both machines very > carefully. What are the specifications on the Mac? the Mac is a dual G5 2GHz, 1.5 GB RAM, 160 GB SATA HDD, it uses ipfw for internet sharing purposes between its eth and its wifi interfaces: # ipfw list 00010 divert 8668 ip from any to any via en0 65535 allow ip from any to any the FreeBSD has 1 GB DDR2 ECC, 2 250GB SATA II HDD (but motherboard controler is on SATA I) and uses pf: # pfctl -s all TRANSLATION RULES: nat on fxp0 inet from 192.168.0.0/24 to any -> (fxp0) round-robin FILTER RULES: scrub in all fragment reassemble block return all block return in log quick proto tcp from to any port = ssh pass quick on lo0 all [ bunch of block in/out and pass in rules applying only on fxp0, the external IF ] pass in inet proto icmp all icmp-type echoreq keep state pass in on em0 inet from 192.168.0.0/24 to any keep state pass out on em0 inet from any to 192.168.0.0/24 keep state [ few pass out rules applying only on fxp0, the external IF ] No queue in use > I averaged 7.5MB/s with sftp and the peak speed on the HTTP > transfer was only 7.1MB/s with Apache 2.2. scp gives me 5.6 MB/s from the Mac to the FreeBSD, but HTTP gives me 20-21 MB/s from the mac (apache 1.3) to the freebsd. Arne's nc trick (server: nc -l 1234 > /dev/null ; client: dd if=/dev/ zero bs=1m | nc serverIP 1234) gives me at best 23.7 MB/s with freebsd as server. About the same with the mac as server. Everything else is still to be tested. thanks, patpro From owner-freebsd-performance@FreeBSD.ORG Sat Jan 13 00:08:14 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id 8714116A494 for ; Sat, 13 Jan 2007 00:08:14 +0000 (UTC) (envelope-from gofp-freebsd-performance@m.gmane.org) Received: from ciao.gmane.org (main.gmane.org [80.91.229.2]) by mx1.freebsd.org (Postfix) with ESMTP id 4268513C442 for ; Sat, 13 Jan 2007 00:08:14 +0000 (UTC) (envelope-from gofp-freebsd-performance@m.gmane.org) Received: from list by ciao.gmane.org with local (Exim 4.43) id 1H5WRA-0007un-Id for freebsd-performance@freebsd.org; Sat, 13 Jan 2007 01:07:56 +0100 Received: from 89-172-39-90.adsl.net.t-com.hr ([89.172.39.90]) by main.gmane.org with esmtp (Gmexim 0.1 (Debian)) id 1AlnuQ-0007hv-00 for ; Sat, 13 Jan 2007 01:07:56 +0100 Received: from ivoras by 89-172-39-90.adsl.net.t-com.hr with local (Gmexim 0.1 (Debian)) id 1AlnuQ-0007hv-00 for ; Sat, 13 Jan 2007 01:07:56 +0100 X-Injected-Via-Gmane: http://gmane.org/ To: freebsd-performance@freebsd.org From: Ivan Voras Date: Sat, 13 Jan 2007 01:07:22 +0100 Lines: 32 Message-ID: References: <465799.40884.qm@web30312.mail.mud.yahoo.com> Mime-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enig6B28B479E0BA7031338ECFC6" X-Complaints-To: usenet@sea.gmane.org X-Gmane-NNTP-Posting-Host: 89-172-39-90.adsl.net.t-com.hr User-Agent: Thunderbird 1.5.0.9 (Windows/20061207) In-Reply-To: X-Enigmail-Version: 0.94.1.2 Sender: news Subject: Re: network perf : em driver ? X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 13 Jan 2007 00:08:14 -0000 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enig6B28B479E0BA7031338ECFC6 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: quoted-printable Chuck Swiger wrote: > Quick testing suggests that an Apache child process accumulates a > similar amount of CPU time transferring large files as scp when using a= n > SSL connection;=20 As expected. Though the original poster didn't mention using SSL, as far as I can see. --------------enig6B28B479E0BA7031338ECFC6 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.4 (MingW32) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFFqCLGldnAQVacBcgRAkb4AKCB4ISUHyNanY5FftV9JQBk6o/WVQCgq9r9 zHhnx2Fz4EuCTktryv+KBPI= =CJAQ -----END PGP SIGNATURE----- --------------enig6B28B479E0BA7031338ECFC6-- From owner-freebsd-performance@FreeBSD.ORG Sat Jan 13 00:15:12 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id A6F8C16A50C for ; Sat, 13 Jan 2007 00:15:12 +0000 (UTC) (envelope-from gofp-freebsd-performance@m.gmane.org) Received: from ciao.gmane.org (main.gmane.org [80.91.229.2]) by mx1.freebsd.org (Postfix) with ESMTP id 369C813C45B for ; Sat, 13 Jan 2007 00:15:12 +0000 (UTC) (envelope-from gofp-freebsd-performance@m.gmane.org) Received: from root by ciao.gmane.org with local (Exim 4.43) id 1H5WY2-0000Jp-12 for freebsd-performance@freebsd.org; Sat, 13 Jan 2007 01:15:02 +0100 Received: from 89-172-39-90.adsl.net.t-com.hr ([89.172.39.90]) by main.gmane.org with esmtp (Gmexim 0.1 (Debian)) id 1AlnuQ-0007hv-00 for ; Sat, 13 Jan 2007 01:15:02 +0100 Received: from ivoras by 89-172-39-90.adsl.net.t-com.hr with local (Gmexim 0.1 (Debian)) id 1AlnuQ-0007hv-00 for ; Sat, 13 Jan 2007 01:15:02 +0100 X-Injected-Via-Gmane: http://gmane.org/ To: freebsd-performance@freebsd.org From: Ivan Voras Date: Sat, 13 Jan 2007 01:12:36 +0100 Lines: 87 Message-ID: References: <465799.40884.qm@web30312.mail.mud.yahoo.com> Mime-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enigA2F99F5C0D6F59185852A690" X-Complaints-To: usenet@sea.gmane.org X-Gmane-NNTP-Posting-Host: 89-172-39-90.adsl.net.t-com.hr User-Agent: Thunderbird 1.5.0.9 (Windows/20061207) In-Reply-To: <465799.40884.qm@web30312.mail.mud.yahoo.com> X-Enigmail-Version: 0.94.1.2 Sender: news Subject: Re: network perf : em driver ? X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 13 Jan 2007 00:15:12 -0000 This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enigA2F99F5C0D6F59185852A690 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: quoted-printable R. B. Riddick wrote: > --- Ivan Voras wrote: >> I've found iperf to be more useful. >> > Soso... Was it slower? Or what? List of options for tcpblast: > tcpblast usage: tcpblast [-4] [-6] destination nblkocks blocksize: 1024 bytes 0 List of options for iperf: > iperf --help Usage: iperf [-s|-c host] [options] iperf [-h|--help] [-v|--version] Client/Server: -f, --format [kmKM] format to report: Kbits, Mbits, KBytes, MBytes= -i, --interval # seconds between periodic bandwidth reports -l, --len #[KM] length of buffer to read or write (default 8 K= B) -m, --print_mss print TCP maximum segment size (MTU - TCP/IP header) -p, --port # server port to listen on/connect to -u, --udp use UDP rather than TCP -w, --window #[KM] TCP window size (socket buffer size) -B, --bind bind to , an interface or multicast addr= ess -C, --compatibility for use with older versions does not sent extra msgs -M, --mss # set TCP maximum segment size (MTU - 40 bytes) -N, --nodelay set TCP no delay, disabling Nagle's Algorithm -V, --IPv6Version Set the domain to IPv6 Server specific: -s, --server run in server mode -U, --single_udp run in single threaded UDP mode -D, --daemon run the server as a daemon Client specific: -b, --bandwidth #[KM] for UDP, bandwidth to send at in bits/sec (default 1 Mbit/sec, implies -u) -c, --client run in client mode, connecting to -d, --dualtest Do a bidirectional test simultaneously -n, --num #[KM] number of bytes to transmit (instead of -t) -r, --tradeoff Do a bidirectional test individually -t, --time # time in seconds to transmit for (default 10 se= cs) -F, --fileinput input the data to be transmitted from a file -I, --stdin input the data to be transmitted from stdin -L, --listenport # port to recieve bidirectional tests back on -P, --parallel # number of parallel client threads to run -T, --ttl # time-to-live, for multicast (default 1) Miscellaneous: -h, --help print this message and quit -v, --version print version information and quit iperf's more useful to me because it can do more and has more tunables. Some of the more useful ones are twiddling Nagle's algorithm and setting TCP window size. --------------enigA2F99F5C0D6F59185852A690 Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.4 (MingW32) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFFqCP6ldnAQVacBcgRAvluAKDZCK6aOs2XPF8vP1MAcdlnbgHLvwCfe52Y GSYka/3KWKeBlNgQGYlhlj0= =ibVl -----END PGP SIGNATURE----- --------------enigA2F99F5C0D6F59185852A690-- From owner-freebsd-performance@FreeBSD.ORG Sat Jan 13 00:26:45 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id 1FFF316A492 for ; Sat, 13 Jan 2007 00:26:45 +0000 (UTC) (envelope-from fbsd-performance@mawer.org) Received: from customer-domains.icp-qv1-irony7.iinet.net.au (customer-domains.icp-qv1-irony7.iinet.net.au [203.59.1.128]) by mx1.freebsd.org (Postfix) with ESMTP id 963C113C468 for ; Sat, 13 Jan 2007 00:26:42 +0000 (UTC) (envelope-from fbsd-performance@mawer.org) Received: from 203-206-173-235.perm.iinet.net.au (HELO [10.24.1.1]) ([203.206.173.235]) by customer-domains.icp-qv1-irony7.iinet.net.au with ESMTP; 13 Jan 2007 07:56:31 +0800 X-IronPort-Anti-Spam-Filtered: true X-IronPort-Anti-Spam-Result: AgAAAIGup0XLzq3r/2dsb2JhbAAN X-IronPort-AV: i="4.13,180,1167580800"; d="scan'208"; a="480304591:sNHT15409532" Message-ID: <45A81FA5.3090701@mawer.org> Date: Sat, 13 Jan 2007 10:54:13 +1100 From: Antony Mawer User-Agent: Thunderbird 1.5.0.9 (Windows/20061207) MIME-Version: 1.0 To: Patrick Proniewski References: <7CA8AE1A-3925-404C-9F69-32AC4FFBB379@patpro.net> In-Reply-To: <7CA8AE1A-3925-404C-9F69-32AC4FFBB379@patpro.net> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: Lucas Holt , freebsd-performance@freebsd.org Subject: Re: network perf : em driver ? X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 13 Jan 2007 00:26:45 -0000 On 13/01/2007 10:08 AM, Patrick Proniewski wrote: > the FreeBSD has 1 GB DDR2 ECC, 2 250GB SATA II HDD (but motherboard > controler is on SATA I) and uses pf: > > # pfctl -s all > TRANSLATION RULES: > nat on fxp0 inet from 192.168.0.0/24 to any -> (fxp0) round-robin > > FILTER RULES: > scrub in all fragment reassemble > block return all > block return in log quick proto tcp from to any port = ssh > pass quick on lo0 all > [ bunch of block in/out and pass in rules applying only on fxp0, the > external IF ] > pass in inet proto icmp all icmp-type echoreq keep state > pass in on em0 inet from 192.168.0.0/24 to any keep state > pass out on em0 inet from any to 192.168.0.0/24 keep state > [ few pass out rules applying only on fxp0, the external IF ] > No queue in use Does disabling pf/clearing out the rules make any difference to the speeds you can achieve? In particular I'm not sure what the performance impact of the "scrub in all" might be. Certainly worth a quick test! --Antony From owner-freebsd-performance@FreeBSD.ORG Sat Jan 13 02:44:37 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id 3072216A40F for ; Sat, 13 Jan 2007 02:44:37 +0000 (UTC) (envelope-from michel@lucenet.com.br) Received: from msrv.matik.com.br (msrv.matik.com.br [200.152.83.14]) by mx1.freebsd.org (Postfix) with ESMTP id B127213C455 for ; Sat, 13 Jan 2007 02:44:33 +0000 (UTC) (envelope-from michel@lucenet.com.br) Received: from webmail.matik.com.br (localhost [127.0.0.1]) by msrv.matik.com.br (8.13.8/8.13.1) with ESMTP id l0D1RrDs076799 for ; Fri, 12 Jan 2007 23:27:53 -0200 (BRST) (envelope-from michel@lucenet.com.br) Received: from 200.152.83.36 (SquirrelMail authenticated user luc.michel) by webmail.matik.com.br with HTTP; Fri, 12 Jan 2007 23:27:53 -0200 (BRST) Message-ID: <64656.200.152.83.36.1168651673.squirrel@webmail.matik.com.br> Date: Fri, 12 Jan 2007 23:27:53 -0200 (BRST) From: "Michel Santos" To: freebsd-performance@freebsd.org User-Agent: SquirrelMail/1.4.9a MIME-Version: 1.0 Content-Type: text/plain;charset=iso-8859-1 Content-Transfer-Encoding: 8bit X-Priority: 3 (Normal) Importance: Normal Subject: diskio low read performance X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 13 Jan 2007 02:44:37 -0000 I hope somebody can help me. I upgraded recently my cache server (squid) 4.11 to 6.2 I notice that my disk read performance seems to be low. I observe it by reading diskio from net-snmp. I started to check this because my old server gave much better results from cache. I see read access as 1/4 - 1/6 of write access. My cache_dirs are full and no swap in use. IOLA is not over 25-30 in peaks is there some special configuration to get better read performance on 6.2? My disks are UW320 10k on Adaptec 29320 and the same as on the 4.11 server. I really dont care about write speed so much so if there are parameters to configure I apreciate to hear about. I checked the following with 1 and 0 but it does not make any difference at all vfs.write_behind vfs.vmiodirenable I get an average of 6-8Mb/s through this server and 3-4 of it is http traffic. I see the object are going into the cache but for some reason it seems they can not be read in time. I tried different newfs -b and -f and actually 1024/4096 seems to be best. I do not cache large files on this machine. Is there something I sgould try? Michel computador é como nem cavalo e mulher mais que montam neles, pior que ficam ... **************************************************** Datacenter Matik http://datacenter.matik.com.br E-Mail e Data Hosting Service para Profissionais. **************************************************** From owner-freebsd-performance@FreeBSD.ORG Sat Jan 13 06:13:17 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id B269516A412 for ; Sat, 13 Jan 2007 06:13:17 +0000 (UTC) (envelope-from markir@paradise.net.nz) Received: from smtp4.clear.net.nz (smtp4.clear.net.nz [203.97.37.64]) by mx1.freebsd.org (Postfix) with ESMTP id 75CFE13C428 for ; Sat, 13 Jan 2007 06:13:17 +0000 (UTC) (envelope-from markir@paradise.net.nz) Received: from [192.168.1.11] (121-72-72-39.dsl.telstraclear.net [121.72.72.39]) by smtp4.clear.net.nz (CLEAR Net Mail) with ESMTP id <0JBS00HJZMM1D930@smtp4.clear.net.nz> for freebsd-performance@freebsd.org; Sat, 13 Jan 2007 19:13:13 +1300 (NZDT) Date: Sat, 13 Jan 2007 19:13:12 +1300 From: Mark Kirkwood In-reply-to: <64656.200.152.83.36.1168651673.squirrel@webmail.matik.com.br> To: Michel Santos Message-id: <45A87878.1050505@paradise.net.nz> MIME-version: 1.0 Content-type: text/plain; charset=ISO-8859-1; format=flowed Content-transfer-encoding: 7bit References: <64656.200.152.83.36.1168651673.squirrel@webmail.matik.com.br> User-Agent: Thunderbird 1.5.0.9 (X11/20061227) Cc: freebsd-performance@freebsd.org Subject: Re: diskio low read performance X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 13 Jan 2007 06:13:17 -0000 Michel Santos wrote: > I hope somebody can help me. I upgraded recently my cache server (squid) > 4.11 to 6.2 > > I notice that my disk read performance seems to be low. I observe it by > reading diskio from net-snmp. I started to check this because my old > server gave much better results from cache. > > I see read access as 1/4 - 1/6 of write access. My cache_dirs are full and > no swap in use. IOLA is not over 25-30 in peaks > > is there some special configuration to get better read performance on 6.2? > My disks are UW320 10k on Adaptec 29320 and the same as on the 4.11 > server. I really dont care about write speed so much so if there are > parameters to configure I apreciate to hear about. > > I checked the following with 1 and 0 but it does not make any difference > at all > > vfs.write_behind > vfs.vmiodirenable > > I get an average of 6-8Mb/s through this server and 3-4 of it is http > traffic. I see the object are going into the cache but for some reason it > seems they can not be read in time. > > I tried different newfs -b and -f and actually 1024/4096 seems to be best. > I do not cache large files on this machine. > > > Is there something I sgould try? > Note sure this will help you, as your files are all small, but try increasing vfs.read_max (say 16 or 32). Cheers Mark From owner-freebsd-performance@FreeBSD.ORG Sat Jan 13 11:53:50 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id C3EBD16A403 for ; Sat, 13 Jan 2007 11:53:50 +0000 (UTC) (envelope-from michel@lucenet.com.br) Received: from msrv.matik.com.br (msrv.matik.com.br [200.152.83.14]) by mx1.freebsd.org (Postfix) with ESMTP id 3453513C442 for ; Sat, 13 Jan 2007 11:53:49 +0000 (UTC) (envelope-from michel@lucenet.com.br) Received: from webmail.matik.com.br (localhost [127.0.0.1]) by msrv.matik.com.br (8.13.8/8.13.1) with ESMTP id l0DBrkOJ017508; Sat, 13 Jan 2007 09:53:47 -0200 (BRST) (envelope-from michel@lucenet.com.br) Received: from 200.152.83.36 (SquirrelMail authenticated user luc.michel) by webmail.matik.com.br with HTTP; Sat, 13 Jan 2007 09:53:47 -0200 (BRST) Message-ID: <63758.200.152.83.36.1168689227.squirrel@webmail.matik.com.br> In-Reply-To: <45A87878.1050505@paradise.net.nz> References: <64656.200.152.83.36.1168651673.squirrel@webmail.matik.com.br> <45A87878.1050505@paradise.net.nz> Date: Sat, 13 Jan 2007 09:53:47 -0200 (BRST) From: "Michel Santos" To: "Mark Kirkwood" User-Agent: SquirrelMail/1.4.9a MIME-Version: 1.0 Content-Type: text/plain;charset=iso-8859-1 Content-Transfer-Encoding: 8bit X-Priority: 3 (Normal) Importance: Normal Cc: freebsd-performance@freebsd.org Subject: Re: diskio low read performance X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 13 Jan 2007 11:53:50 -0000 Mark Kirkwood disse na ultima mensagem: > Michel Santos wrote: >> I hope somebody can help me. I upgraded recently my cache server (squid) >> 4.11 to 6.2 >> >> I notice that my disk read performance seems to be low. I observe it by >> reading diskio from net-snmp. I started to check this because my old >> server gave much better results from cache. >> >> I see read access as 1/4 - 1/6 of write access. My cache_dirs are full >> and >> no swap in use. IOLA is not over 25-30 in peaks >> >> is there some special configuration to get better read performance on >> 6.2? >> My disks are UW320 10k on Adaptec 29320 and the same as on the 4.11 >> server. I really dont care about write speed so much so if there are >> parameters to configure I apreciate to hear about. >> >> I checked the following with 1 and 0 but it does not make any difference >> at all >> >> vfs.write_behind >> vfs.vmiodirenable >> >> I get an average of 6-8Mb/s through this server and 3-4 of it is http >> traffic. I see the object are going into the cache but for some reason >> it >> seems they can not be read in time. >> >> I tried different newfs -b and -f and actually 1024/4096 seems to be >> best. >> I do not cache large files on this machine. >> >> >> Is there something I sgould try? >> > > Note sure this will help you, as your files are all small, but try > increasing vfs.read_max (say 16 or 32). > I forgot to say that I tried it already. Even if it gave me no improvement I have it in 16 at this time together with a higher vfs.ufs.dirhash_maxmem value. Sincerley, any of the configuration changes I did gave me absolutely nothing in relationship to the disk read access performance. That is disappointing. Should I go back and try ufs1 perhaps? Or is it that squid does not work well on 6.2? I use diskd but in order to check if there is a SHM issue I tried ufs and aufs, but also no difference at all. On 6.2 I do not even get close to 50% of 4.11 disk read performance. thank you Michel computador é como nem cavalo e mulher mais que montam neles, pior que ficam ... **************************************************** Datacenter Matik http://datacenter.matik.com.br E-Mail e Data Hosting Service para Profissionais. **************************************************** From owner-freebsd-performance@FreeBSD.ORG Sat Jan 13 13:36:23 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id B06C216A412 for ; Sat, 13 Jan 2007 13:36:23 +0000 (UTC) (envelope-from patpro@patpro.net) Received: from smtp1-g19.free.fr (smtp1-g19.free.fr [212.27.42.27]) by mx1.freebsd.org (Postfix) with ESMTP id 7D97313C442 for ; Sat, 13 Jan 2007 13:36:23 +0000 (UTC) (envelope-from patpro@patpro.net) Received: from [192.168.0.2] (boleskine.patpro.net [82.235.12.223]) by smtp1-g19.free.fr (Postfix) with ESMTP id C4FB69B6A7 for ; Sat, 13 Jan 2007 14:36:22 +0100 (CET) Mime-Version: 1.0 (Apple Message framework v752.2) In-Reply-To: <45A81FA5.3090701@mawer.org> References: <7CA8AE1A-3925-404C-9F69-32AC4FFBB379@patpro.net> <45A81FA5.3090701@mawer.org> Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: Content-Transfer-Encoding: 7bit From: Patrick Proniewski Date: Sat, 13 Jan 2007 14:36:21 +0100 To: freebsd-performance@freebsd.org X-Mailer: Apple Mail (2.752.2) Subject: Re: network perf : em driver ? [solved] X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 13 Jan 2007 13:36:23 -0000 Thank you all for your help. I've finally found the bottleneck: the onboard PowerMac ethernet port. After verification, changing network cables was not an option, I've used on this connection a cat. 7 network cable. I've connected the em0 port of the freebsd box on a gigabit PCI-X NIC I have in the Mac, and given a second try to the "dd | nc" client/ server bench: 103 MB/s. This is far better ! Same setup with apache 1.3 on the mac, wget on the Freebsd: about 40 MB/s (350 MB file) The PCI-X NIC has flow-control enabled: media: autoselect (1000baseT ) status: active I'll check later the influence of this parameter. Again, thank you all for your replies patpro From owner-freebsd-performance@FreeBSD.ORG Sat Jan 13 14:31:51 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id E983B16A407 for ; Sat, 13 Jan 2007 14:31:51 +0000 (UTC) (envelope-from arne_woerner@yahoo.com) Received: from web30301.mail.mud.yahoo.com (web30301.mail.mud.yahoo.com [209.191.69.63]) by mx1.freebsd.org (Postfix) with SMTP id 9F41413C43E for ; Sat, 13 Jan 2007 14:31:51 +0000 (UTC) (envelope-from arne_woerner@yahoo.com) Received: (qmail 50338 invoked by uid 60001); 13 Jan 2007 14:31:51 -0000 DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.com; h=X-YMail-OSG:Received:Date:From:Subject:To:In-Reply-To:MIME-Version:Content-Type:Content-Transfer-Encoding:Message-ID; b=fS1jx6zRz4P7MZwPlquIwv+GJdYCudHV0+s870LwjMRcdH7/7aWeaqwQaw5VWW6sV1f/Mlq1JdNwC2OuR1okugQURODPQSLGasYydYfzfvQSb9jo/6/RS+GcDqpKG+82k/gVODRG9fq2sS4C5GLwRqs/wb8eA3bkGI9jEqYxAQ8=; X-YMail-OSG: ZyWWMCwVM1n5GV8xlWiMMMpTdaM3qzsk4KOVzMDm Received: from [213.54.176.27] by web30301.mail.mud.yahoo.com via HTTP; Sat, 13 Jan 2007 06:31:50 PST Date: Sat, 13 Jan 2007 06:31:50 -0800 (PST) From: "R. B. Riddick" To: Patrick Proniewski , freebsd-performance@freebsd.org In-Reply-To: MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit Message-ID: <988678.50280.qm@web30301.mail.mud.yahoo.com> Cc: Subject: Re: network perf : em driver ? [solved] X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 13 Jan 2007 14:31:52 -0000 --- Patrick Proniewski wrote: > I've connected the em0 port of the freebsd box on a gigabit PCI-X NIC > I have in the Mac, and given a second try to the "dd | nc" client/ > server bench: 103 MB/s. This is far better ! > Same setup with apache 1.3 on the mac, wget on the Freebsd: about 40 > MB/s (350 MB file) > May I see the results of thttpd in comparison to apache (just for the file)? :-) And I would like to direct ur attention to this thread on freebsd-geom@ (that could make ur disk-bottleneck go away; cave: disk bus contention): http://docs.freebsd.org/cgi/mid.cgi?20070113004728.GQ2616 (esp.: http://docs.freebsd.org/cgi/mid.cgi?45A8B722.7020302) -Arne ____________________________________________________________________________________ No need to miss a message. Get email on-the-go with Yahoo! Mail for Mobile. Get started. http://mobile.yahoo.com/mail From owner-freebsd-performance@FreeBSD.ORG Sat Jan 13 16:07:35 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id 4529416A415 for ; Sat, 13 Jan 2007 16:07:35 +0000 (UTC) (envelope-from patpro@patpro.net) Received: from smtp8-g19.free.fr (smtp8-g19.free.fr [212.27.42.65]) by mx1.freebsd.org (Postfix) with ESMTP id 13B1D13C4A8 for ; Sat, 13 Jan 2007 16:07:34 +0000 (UTC) (envelope-from patpro@patpro.net) Received: from [192.168.0.2] (boleskine.patpro.net [82.235.12.223]) by smtp8-g19.free.fr (Postfix) with ESMTP id 227A65578; Sat, 13 Jan 2007 17:07:33 +0100 (CET) In-Reply-To: <988678.50280.qm@web30301.mail.mud.yahoo.com> References: <988678.50280.qm@web30301.mail.mud.yahoo.com> Mime-Version: 1.0 (Apple Message framework v752.2) Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed Message-Id: Content-Transfer-Encoding: 7bit From: Patrick Proniewski Date: Sat, 13 Jan 2007 17:07:32 +0100 To: "R. B. Riddick" X-Mailer: Apple Mail (2.752.2) Cc: freebsd-performance@freebsd.org Subject: Re: network perf : em driver ? [solved] X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 13 Jan 2007 16:07:35 -0000 On 13 janv. 2007, at 15:31, R. B. Riddick wrote: > --- Patrick Proniewski wrote: >> I've connected the em0 port of the freebsd box on a gigabit PCI-X NIC >> I have in the Mac, and given a second try to the "dd | nc" client/ >> server bench: 103 MB/s. This is far better ! >> Same setup with apache 1.3 on the mac, wget on the Freebsd: about 40 >> MB/s (350 MB file) >> > May I see the results of thttpd in comparison to apache (just for > the file)? > :-) I made a quick&dirty compilation of thttpd on the Mac, and I've got about 39 MB/s, so quite the same. I made further tests, and I've discovered that the real culprit is ipfw/natd on the Mac. When I share LAN/Internet access on the Mac from eth to airport wifi, ipfw/natd are configured in such a way that my eth network crawls. When I've changed NICs (between onboard and PCI-X) the internet sharing was still configured to use onboard eth. But if I reconfigure the internet sharing on the Mac to use the PCI-X NIC, the speed lowers to 21 MB/s instead of 40. I'll have to find a good wifi card for the freebsd box, so that the Mac won't have to be a gateway-behind-the-gateway. > And I would like to direct ur attention to this thread on freebsd- > geom@ (that > could make ur disk-bottleneck go away; cave: disk bus contention): > http://docs.freebsd.org/cgi/mid.cgi?20070113004728.GQ2616 > (esp.: http://docs.freebsd.org/cgi/mid.cgi?45A8B722.7020302) I'll take a look, thanks ! patpro From owner-freebsd-performance@FreeBSD.ORG Sat Jan 13 16:42:35 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id 2806216A403 for ; Sat, 13 Jan 2007 16:42:35 +0000 (UTC) (envelope-from kris@obsecurity.org) Received: from elvis.mu.org (elvis.mu.org [192.203.228.196]) by mx1.freebsd.org (Postfix) with ESMTP id 1504B13C45A for ; Sat, 13 Jan 2007 16:42:35 +0000 (UTC) (envelope-from kris@obsecurity.org) Received: from obsecurity.dyndns.org (elvis.mu.org [192.203.228.196]) by elvis.mu.org (Postfix) with ESMTP id EA2421A4D82; Sat, 13 Jan 2007 08:42:34 -0800 (PST) Received: by obsecurity.dyndns.org (Postfix, from userid 1000) id E3529512A1; Sat, 13 Jan 2007 11:42:32 -0500 (EST) Date: Sat, 13 Jan 2007 11:42:32 -0500 From: Kris Kennaway To: Michel Santos Message-ID: <20070113164232.GA34348@xor.obsecurity.org> References: <64656.200.152.83.36.1168651673.squirrel@webmail.matik.com.br> <45A87878.1050505@paradise.net.nz> <63758.200.152.83.36.1168689227.squirrel@webmail.matik.com.br> Mime-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="pf9I7BMVVzbSWLtt" Content-Disposition: inline In-Reply-To: <63758.200.152.83.36.1168689227.squirrel@webmail.matik.com.br> User-Agent: Mutt/1.4.2.2i Cc: freebsd-performance@freebsd.org, Mark Kirkwood Subject: Re: diskio low read performance X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 13 Jan 2007 16:42:35 -0000 --pf9I7BMVVzbSWLtt Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable On Sat, Jan 13, 2007 at 09:53:47AM -0200, Michel Santos wrote: > I forgot to say that I tried it already. Even if it gave me no improvement > I have it in 16 at this time together with a higher vfs.ufs.dirhash_maxmem > value. >=20 > Sincerley, any of the configuration changes I did gave me absolutely > nothing in relationship to the disk read access performance. That is > disappointing. >=20 > Should I go back and try ufs1 perhaps? Or is it that squid does not work > well on 6.2? Is it the same version of squid, same configuration, etc? Kris --pf9I7BMVVzbSWLtt Content-Type: application/pgp-signature Content-Disposition: inline -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (FreeBSD) iD8DBQFFqQv4Wry0BWjoQKURAlkNAJ9Hnieo6LKxkLVt94YeS6fyWQGyWACfVRN7 oIzYaxYr46/UhzV+NvGtVfo= =2m4d -----END PGP SIGNATURE----- --pf9I7BMVVzbSWLtt-- From owner-freebsd-performance@FreeBSD.ORG Sat Jan 13 17:41:24 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id 9C6F816A407 for ; Sat, 13 Jan 2007 17:41:24 +0000 (UTC) (envelope-from michel@lucenet.com.br) Received: from msrv.matik.com.br (msrv.matik.com.br [200.152.83.14]) by mx1.freebsd.org (Postfix) with ESMTP id 1977713C428 for ; Sat, 13 Jan 2007 17:41:23 +0000 (UTC) (envelope-from michel@lucenet.com.br) Received: from webmail.matik.com.br (localhost [127.0.0.1]) by msrv.matik.com.br (8.13.8/8.13.1) with ESMTP id l0DHfK1l043621; Sat, 13 Jan 2007 15:41:20 -0200 (BRST) (envelope-from michel@lucenet.com.br) Received: from 200.152.83.36 (SquirrelMail authenticated user luc.michel) by webmail.matik.com.br with HTTP; Sat, 13 Jan 2007 15:41:21 -0200 (BRST) Message-ID: <64857.200.152.83.36.1168710081.squirrel@webmail.matik.com.br> In-Reply-To: <20070113164232.GA34348@xor.obsecurity.org> References: <64656.200.152.83.36.1168651673.squirrel@webmail.matik.com.br> <45A87878.1050505@paradise.net.nz> <63758.200.152.83.36.1168689227.squirrel@webmail.matik.com.br> <20070113164232.GA34348@xor.obsecurity.org> Date: Sat, 13 Jan 2007 15:41:21 -0200 (BRST) From: "Michel Santos" To: "Kris Kennaway" User-Agent: SquirrelMail/1.4.9a MIME-Version: 1.0 Content-Type: text/plain;charset=iso-8859-1 Content-Transfer-Encoding: 8bit X-Priority: 3 (Normal) Importance: Normal Cc: freebsd-performance@freebsd.org, Mark Kirkwood Subject: Re: diskio low read performance X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 13 Jan 2007 17:41:24 -0000 Kris Kennaway disse na ultima mensagem: > On Sat, Jan 13, 2007 at 09:53:47AM -0200, Michel Santos wrote: > >> I forgot to say that I tried it already. Even if it gave me no >> improvement >> I have it in 16 at this time together with a higher >> vfs.ufs.dirhash_maxmem >> value. >> >> Sincerley, any of the configuration changes I did gave me absolutely >> nothing in relationship to the disk read access performance. That is >> disappointing. >> >> Should I go back and try ufs1 perhaps? Or is it that squid does not work >> well on 6.2? > > Is it the same version of squid, same configuration, etc? > > Kris > Yes, if you used to squid I am running the last 2.5-Stable14 version which run best on 4.11 Michel computador é como nem cavalo e mulher mais que montam neles, pior que ficam ... **************************************************** Datacenter Matik http://datacenter.matik.com.br E-Mail e Data Hosting Service para Profissionais. **************************************************** From owner-freebsd-performance@FreeBSD.ORG Sat Jan 13 18:00:41 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id 333E316A403 for ; Sat, 13 Jan 2007 18:00:41 +0000 (UTC) (envelope-from kris@obsecurity.org) Received: from elvis.mu.org (elvis.mu.org [192.203.228.196]) by mx1.freebsd.org (Postfix) with ESMTP id 1D11613C44B for ; Sat, 13 Jan 2007 18:00:41 +0000 (UTC) (envelope-from kris@obsecurity.org) Received: from obsecurity.dyndns.org (elvis.mu.org [192.203.228.196]) by elvis.mu.org (Postfix) with ESMTP id B72AB1A4D81; Sat, 13 Jan 2007 10:00:40 -0800 (PST) Received: by obsecurity.dyndns.org (Postfix, from userid 1000) id 16B5E5138A; Sat, 13 Jan 2007 13:00:36 -0500 (EST) Date: Sat, 13 Jan 2007 13:00:36 -0500 From: Kris Kennaway To: Michel Santos Message-ID: <20070113180036.GA64359@xor.obsecurity.org> References: <64656.200.152.83.36.1168651673.squirrel@webmail.matik.com.br> <45A87878.1050505@paradise.net.nz> <63758.200.152.83.36.1168689227.squirrel@webmail.matik.com.br> <20070113164232.GA34348@xor.obsecurity.org> <64857.200.152.83.36.1168710081.squirrel@webmail.matik.com.br> Mime-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="ikeVEW9yuYc//A+q" Content-Disposition: inline In-Reply-To: <64857.200.152.83.36.1168710081.squirrel@webmail.matik.com.br> User-Agent: Mutt/1.4.2.2i Cc: freebsd-performance@freebsd.org, Mark Kirkwood , Kris Kennaway Subject: Re: diskio low read performance X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 13 Jan 2007 18:00:41 -0000 --ikeVEW9yuYc//A+q Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable On Sat, Jan 13, 2007 at 03:41:21PM -0200, Michel Santos wrote: >=20 > Kris Kennaway disse na ultima mensagem: > > On Sat, Jan 13, 2007 at 09:53:47AM -0200, Michel Santos wrote: > > > >> I forgot to say that I tried it already. Even if it gave me no > >> improvement > >> I have it in 16 at this time together with a higher > >> vfs.ufs.dirhash_maxmem > >> value. > >> > >> Sincerley, any of the configuration changes I did gave me absolutely > >> nothing in relationship to the disk read access performance. That is > >> disappointing. > >> > >> Should I go back and try ufs1 perhaps? Or is it that squid does not wo= rk > >> well on 6.2? > > > > Is it the same version of squid, same configuration, etc? > > > > Kris > > >=20 > Yes, if you used to squid I am running the last 2.5-Stable14 version which > run best on 4.11 OK, please provide further details of your system configuration (dmesg, kernel config). Kris --ikeVEW9yuYc//A+q Content-Type: application/pgp-signature Content-Disposition: inline -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (FreeBSD) iD8DBQFFqR5EWry0BWjoQKURAr2AAJ0Z8hCn4LIvUezg5xbjLdGPrBtJfQCdHt0q R1M3fcrcq4kAhs3w7q8KLGo= =oloW -----END PGP SIGNATURE----- --ikeVEW9yuYc//A+q-- From owner-freebsd-performance@FreeBSD.ORG Sat Jan 13 18:58:08 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id 636E216A407 for ; Sat, 13 Jan 2007 18:58:08 +0000 (UTC) (envelope-from michel@lucenet.com.br) Received: from msrv.matik.com.br (msrv.matik.com.br [200.152.83.14]) by mx1.freebsd.org (Postfix) with ESMTP id C1BAE13C458 for ; Sat, 13 Jan 2007 18:58:07 +0000 (UTC) (envelope-from michel@lucenet.com.br) Received: from webmail.matik.com.br (localhost [127.0.0.1]) by msrv.matik.com.br (8.13.8/8.13.1) with ESMTP id l0DIw0vQ048631; Sat, 13 Jan 2007 16:58:05 -0200 (BRST) (envelope-from michel@lucenet.com.br) Received: from 200.152.83.36 (SquirrelMail authenticated user luc.michel) by webmail.matik.com.br with HTTP; Sat, 13 Jan 2007 16:58:06 -0200 (BRST) Message-ID: <60639.200.152.83.36.1168714686.squirrel@webmail.matik.com.br> In-Reply-To: <20070113180036.GA64359@xor.obsecurity.org> References: <64656.200.152.83.36.1168651673.squirrel@webmail.matik.com.br> <45A87878.1050505@paradise.net.nz> <63758.200.152.83.36.1168689227.squirrel@webmail.matik.com.br> <20070113164232.GA34348@xor.obsecurity.org> <64857.200.152.83.36.1168710081.squirrel@webmail.matik.com.br> <20070113180036.GA64359@xor.obsecurity.org> Date: Sat, 13 Jan 2007 16:58:06 -0200 (BRST) From: "Michel Santos" To: "Kris Kennaway" User-Agent: SquirrelMail/1.4.9a MIME-Version: 1.0 Content-Type: text/plain;charset=iso-8859-1 Content-Transfer-Encoding: 8bit X-Priority: 3 (Normal) Importance: Normal Cc: freebsd-performance@freebsd.org, Mark Kirkwood , Kris Kennaway Subject: Re: diskio low read performance X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 13 Jan 2007 18:58:08 -0000 Kris Kennaway disse na ultima mensagem: > On Sat, Jan 13, 2007 at 03:41:21PM -0200, Michel Santos wrote: >> >> Kris Kennaway disse na ultima mensagem: >> > On Sat, Jan 13, 2007 at 09:53:47AM -0200, Michel Santos wrote: >> > >> >> I forgot to say that I tried it already. Even if it gave me no >> >> improvement >> >> I have it in 16 at this time together with a higher >> >> vfs.ufs.dirhash_maxmem >> >> value. >> >> >> >> Sincerley, any of the configuration changes I did gave me absolutely >> >> nothing in relationship to the disk read access performance. That is >> >> disappointing. >> >> >> >> Should I go back and try ufs1 perhaps? Or is it that squid does not >> work >> >> well on 6.2? >> > >> > Is it the same version of squid, same configuration, etc? >> > >> > Kris >> > >> >> Yes, if you used to squid I am running the last 2.5-Stable14 version >> which >> run best on 4.11 > > OK, please provide further details of your system configuration > (dmesg, kernel config). > I have two server, the main server is a Supermicro Dualcore Dual Opteron and the backup is a Athlon64 X2, both with 4GB The disks are the same, only the onboard SCSI is Adaptec and the other is LSI. Funny is that I have no difference regarding the disk read performance wether I use the Opteron machine or the other dmesg Opteron http://suporte.lucenet.com.br/ms/dmesg.sm dmesg X2 http://suporte.lucenet.com.br/ms/dmesg.x2 kernel config http://suporte.lucenet.com.br/ms/kernel62 thank you Michel computador é como nem cavalo e mulher mais que montam neles, pior que ficam ... **************************************************** Datacenter Matik http://datacenter.matik.com.br E-Mail e Data Hosting Service para Profissionais. **************************************************** From owner-freebsd-performance@FreeBSD.ORG Sat Jan 13 19:04:49 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id D6C5116A415 for ; Sat, 13 Jan 2007 19:04:49 +0000 (UTC) (envelope-from kris@obsecurity.org) Received: from elvis.mu.org (elvis.mu.org [192.203.228.196]) by mx1.freebsd.org (Postfix) with ESMTP id BEE7313C458 for ; Sat, 13 Jan 2007 19:04:49 +0000 (UTC) (envelope-from kris@obsecurity.org) Received: from obsecurity.dyndns.org (elvis.mu.org [192.203.228.196]) by elvis.mu.org (Postfix) with ESMTP id A02251A4D82; Sat, 13 Jan 2007 11:04:49 -0800 (PST) Received: by obsecurity.dyndns.org (Postfix, from userid 1000) id 7C46D51593; Sat, 13 Jan 2007 14:04:47 -0500 (EST) Date: Sat, 13 Jan 2007 14:04:47 -0500 From: Kris Kennaway To: Michel Santos Message-ID: <20070113190447.GA65571@xor.obsecurity.org> References: <64656.200.152.83.36.1168651673.squirrel@webmail.matik.com.br> <45A87878.1050505@paradise.net.nz> <63758.200.152.83.36.1168689227.squirrel@webmail.matik.com.br> <20070113164232.GA34348@xor.obsecurity.org> <64857.200.152.83.36.1168710081.squirrel@webmail.matik.com.br> <20070113180036.GA64359@xor.obsecurity.org> <60639.200.152.83.36.1168714686.squirrel@webmail.matik.com.br> Mime-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="BOKacYhQ+x31HxR3" Content-Disposition: inline In-Reply-To: <60639.200.152.83.36.1168714686.squirrel@webmail.matik.com.br> User-Agent: Mutt/1.4.2.2i Cc: freebsd-performance@freebsd.org, Mark Kirkwood , Kris Kennaway Subject: Re: diskio low read performance X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 13 Jan 2007 19:04:49 -0000 --BOKacYhQ+x31HxR3 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable On Sat, Jan 13, 2007 at 04:58:06PM -0200, Michel Santos wrote: > I have two server, the main server is a Supermicro Dualcore Dual Opteron > and the backup is a Athlon64 X2, both with 4GB >=20 > The disks are the same, only the onboard SCSI is Adaptec and the other is > LSI. Funny is that I have no difference regarding the disk read > performance wether I use the Opteron machine or the other >=20 > dmesg Opteron > http://suporte.lucenet.com.br/ms/dmesg.sm >=20 > dmesg X2 > http://suporte.lucenet.com.br/ms/dmesg.x2 >=20 > kernel config > http://suporte.lucenet.com.br/ms/kernel62 options SCHED_ULE # ULE scheduler =46rom the NOTES file from where you copied this: # SCHED_ULE is a new scheduler that has been designed for SMP and has some # advantages for UP as well. It is intended to replace the 4BSD scheduler # over time. NOTE: SCHED_ULE is currently considered experimental and is # not recommended for production use at this time. When investigating problems with your system, your very first step should be to revert the use of code marked "experimental" and "not recommended for production use" ;-) Kris --BOKacYhQ+x31HxR3 Content-Type: application/pgp-signature Content-Disposition: inline -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (FreeBSD) iD8DBQFFqS1PWry0BWjoQKURAqjIAKDySjCjqjr73WibSWfVBuLNcfEpMgCeLsmO Qv4AL2FMo5EI3t1ErRZW/MU= =jCfx -----END PGP SIGNATURE----- --BOKacYhQ+x31HxR3-- From owner-freebsd-performance@FreeBSD.ORG Sat Jan 13 19:23:41 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id 81ECC16A403 for ; Sat, 13 Jan 2007 19:23:41 +0000 (UTC) (envelope-from michel@lucenet.com.br) Received: from msrv.matik.com.br (msrv.matik.com.br [200.152.83.14]) by mx1.freebsd.org (Postfix) with ESMTP id E003113C44B for ; Sat, 13 Jan 2007 19:23:40 +0000 (UTC) (envelope-from michel@lucenet.com.br) Received: from webmail.matik.com.br (localhost [127.0.0.1]) by msrv.matik.com.br (8.13.8/8.13.1) with ESMTP id l0DJNa1V050287; Sat, 13 Jan 2007 17:23:37 -0200 (BRST) (envelope-from michel@lucenet.com.br) Received: from 200.152.83.36 (SquirrelMail authenticated user luc.michel) by webmail.matik.com.br with HTTP; Sat, 13 Jan 2007 17:23:39 -0200 (BRST) Message-ID: <64716.200.152.83.36.1168716219.squirrel@webmail.matik.com.br> In-Reply-To: <20070113190447.GA65571@xor.obsecurity.org> References: <64656.200.152.83.36.1168651673.squirrel@webmail.matik.com.br> <45A87878.1050505@paradise.net.nz> <63758.200.152.83.36.1168689227.squirrel@webmail.matik.com.br> <20070113164232.GA34348@xor.obsecurity.org> <64857.200.152.83.36.1168710081.squirrel@webmail.matik.com.br> <20070113180036.GA64359@xor.obsecurity.org> <60639.200.152.83.36.1168714686.squirrel@webmail.matik.com.br> <20070113190447.GA65571@xor.obsecurity.org> Date: Sat, 13 Jan 2007 17:23:39 -0200 (BRST) From: "Michel Santos" To: "Kris Kennaway" User-Agent: SquirrelMail/1.4.9a MIME-Version: 1.0 Content-Type: text/plain;charset=iso-8859-1 Content-Transfer-Encoding: 8bit X-Priority: 3 (Normal) Importance: Normal Cc: freebsd-performance@freebsd.org, Mark Kirkwood Subject: Re: diskio low read performance X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 13 Jan 2007 19:23:41 -0000 Kris Kennaway disse na ultima mensagem: > On Sat, Jan 13, 2007 at 04:58:06PM -0200, Michel Santos wrote: > >> I have two server, the main server is a Supermicro Dualcore Dual Opteron >> and the backup is a Athlon64 X2, both with 4GB >> >> The disks are the same, only the onboard SCSI is Adaptec and the other >> is >> LSI. Funny is that I have no difference regarding the disk read >> performance wether I use the Opteron machine or the other >> >> dmesg Opteron >> http://suporte.lucenet.com.br/ms/dmesg.sm >> >> dmesg X2 >> http://suporte.lucenet.com.br/ms/dmesg.x2 >> >> kernel config >> http://suporte.lucenet.com.br/ms/kernel62 > > options SCHED_ULE # ULE scheduler > > From the NOTES file from where you copied this: > > # SCHED_ULE is a new scheduler that has been designed for SMP and has some > # advantages for UP as well. It is intended to replace the 4BSD scheduler > # over time. NOTE: SCHED_ULE is currently considered experimental and is > # not recommended for production use at this time. > > When investigating problems with your system, your very first step > should be to revert the use of code marked "experimental" and "not > recommended for production use" ;-) > I am running both (on at a time of course :) ), now for six month or so, ULE is giving me better overall performance, either with or w/o polling. I mean network performance. I have net.isr.enable=1 and net.inet.ip.fastforwarding=1, this way I do get the same network performance I had on the 4.11. I mean I have no problem here. But also I checked the ULE/BSD against my particular problem and there is no difference at all. I get no acceptable disk read performance when comparing what I had with 4.11, wether with ULE or with 4BSD Michel computador é como nem cavalo e mulher mais que montam neles, pior que ficam ... **************************************************** Datacenter Matik http://datacenter.matik.com.br E-Mail e Data Hosting Service para Profissionais. **************************************************** From owner-freebsd-performance@FreeBSD.ORG Sat Jan 13 19:41:58 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id 26DB316A47B for ; Sat, 13 Jan 2007 19:41:58 +0000 (UTC) (envelope-from kris@obsecurity.org) Received: from elvis.mu.org (elvis.mu.org [192.203.228.196]) by mx1.freebsd.org (Postfix) with ESMTP id 09F6313C455 for ; Sat, 13 Jan 2007 19:41:58 +0000 (UTC) (envelope-from kris@obsecurity.org) Received: from obsecurity.dyndns.org (elvis.mu.org [192.203.228.196]) by elvis.mu.org (Postfix) with ESMTP id 819B61A4D82; Sat, 13 Jan 2007 11:41:57 -0800 (PST) Received: by obsecurity.dyndns.org (Postfix, from userid 1000) id E282851593; Sat, 13 Jan 2007 14:41:54 -0500 (EST) Date: Sat, 13 Jan 2007 14:41:54 -0500 From: Kris Kennaway To: Michel Santos Message-ID: <20070113194154.GA65864@xor.obsecurity.org> References: <64656.200.152.83.36.1168651673.squirrel@webmail.matik.com.br> <45A87878.1050505@paradise.net.nz> <63758.200.152.83.36.1168689227.squirrel@webmail.matik.com.br> <20070113164232.GA34348@xor.obsecurity.org> <64857.200.152.83.36.1168710081.squirrel@webmail.matik.com.br> <20070113180036.GA64359@xor.obsecurity.org> <60639.200.152.83.36.1168714686.squirrel@webmail.matik.com.br> <20070113190447.GA65571@xor.obsecurity.org> <64716.200.152.83.36.1168716219.squirrel@webmail.matik.com.br> Mime-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="HlL+5n6rz5pIUxbD" Content-Disposition: inline In-Reply-To: <64716.200.152.83.36.1168716219.squirrel@webmail.matik.com.br> User-Agent: Mutt/1.4.2.2i Cc: freebsd-performance@freebsd.org, Mark Kirkwood , Kris Kennaway Subject: Re: diskio low read performance X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 13 Jan 2007 19:41:58 -0000 --HlL+5n6rz5pIUxbD Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable On Sat, Jan 13, 2007 at 05:23:39PM -0200, Michel Santos wrote: > >> kernel config > >> http://suporte.lucenet.com.br/ms/kernel62 > > > > options SCHED_ULE # ULE scheduler > > > > From the NOTES file from where you copied this: > > > > # SCHED_ULE is a new scheduler that has been designed for SMP and has s= ome > > # advantages for UP as well. It is intended to replace the 4BSD schedu= ler > > # over time. NOTE: SCHED_ULE is currently considered experimental and = is > > # not recommended for production use at this time. > > > > When investigating problems with your system, your very first step > > should be to revert the use of code marked "experimental" and "not > > recommended for production use" ;-) > > >=20 > I am running both (on at a time of course :) ), now for six month or so, > ULE is giving me better overall performance, either with or w/o polling. I > mean network performance. I have net.isr.enable=3D1 and > net.inet.ip.fastforwarding=3D1, > this way I do get the same network performance I had on the 4.11. I mean I > have no problem here. >=20 > But also I checked the ULE/BSD against my particular problem and there is > no difference at all. I get no acceptable disk read performance when > comparing what I had with 4.11, wether with ULE or with 4BSD Be very careful, because I've personally measured severe disk I/O penalties with ULE on SMP hardware. In fact in my testing ULE gives worse SMP performance under load across the board compared to 4BSD (it's only faster for the lightest of workloads). If you're absolutely certain that ULE is not to blame (and want to continue to take the risk of other performance and stability problems down the line), that basically leaves something to do with the scsi driver and/or its interaction with your hardware as the probable cause. I don't know enough about this particular hardware to comment further though. Some other comments about your config that are not likely to be relevant for your I/O problem: * it's not recommended to use an explicit maxusers value unless you know that you need it; modern versions of FreeBSD auto-tune with maxusers 0 which is the recommended configuration. * adaptive mutexes are usually a win so it's a bit unusual that you have disabled them, but I assume you have tested this. * Dunno about the AUTO_EOI_1 option, I don't think you even have this hardware on your system (device atpic didn't probe in your dmesg). * HZ=3D1000 is superfluous since it is the default but may or may not help. In some workloads the increased overhead relative to the old default of HZ=3D100 gives a performance loss. Maybe it helps with polling though, I dunno. Kris --HlL+5n6rz5pIUxbD Content-Type: application/pgp-signature Content-Disposition: inline -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (FreeBSD) iD8DBQFFqTYCWry0BWjoQKURAkXNAKCXT3Xgg2Uph0U7H48XI29uy5yFkACfWJHm /uNORKz65sTB4zg65tHW5fw= =kUqf -----END PGP SIGNATURE----- --HlL+5n6rz5pIUxbD-- From owner-freebsd-performance@FreeBSD.ORG Sat Jan 13 20:03:20 2007 Return-Path: X-Original-To: freebsd-performance@freebsd.org Delivered-To: freebsd-performance@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id 35B1116A492 for ; Sat, 13 Jan 2007 20:03:20 +0000 (UTC) (envelope-from michel@lucenet.com.br) Received: from msrv.matik.com.br (msrv.matik.com.br [200.152.83.14]) by mx1.freebsd.org (Postfix) with ESMTP id A057113C458 for ; Sat, 13 Jan 2007 20:03:19 +0000 (UTC) (envelope-from michel@lucenet.com.br) Received: from webmail.matik.com.br (localhost [127.0.0.1]) by msrv.matik.com.br (8.13.8/8.13.1) with ESMTP id l0DK3HEF052964; Sat, 13 Jan 2007 18:03:18 -0200 (BRST) (envelope-from michel@lucenet.com.br) Received: from 200.152.83.36 (SquirrelMail authenticated user luc.michel) by webmail.matik.com.br with HTTP; Sat, 13 Jan 2007 18:03:18 -0200 (BRST) Message-ID: <53057.200.152.83.36.1168718598.squirrel@webmail.matik.com.br> In-Reply-To: <20070113194154.GA65864@xor.obsecurity.org> References: <64656.200.152.83.36.1168651673.squirrel@webmail.matik.com.br> <45A87878.1050505@paradise.net.nz> <63758.200.152.83.36.1168689227.squirrel@webmail.matik.com.br> <20070113164232.GA34348@xor.obsecurity.org> <64857.200.152.83.36.1168710081.squirrel@webmail.matik.com.br> <20070113180036.GA64359@xor.obsecurity.org> <60639.200.152.83.36.1168714686.squirrel@webmail.matik.com.br> <20070113190447.GA65571@xor.obsecurity.org> <64716.200.152.83.36.1168716219.squirrel@webmail.matik.com.br> <20070113194154.GA65864@xor.obsecurity.org> Date: Sat, 13 Jan 2007 18:03:18 -0200 (BRST) From: "Michel Santos" To: "Kris Kennaway" User-Agent: SquirrelMail/1.4.9a MIME-Version: 1.0 Content-Type: text/plain;charset=iso-8859-1 Content-Transfer-Encoding: 8bit X-Priority: 3 (Normal) Importance: Normal Cc: freebsd-performance@freebsd.org, Mark Kirkwood Subject: Re: diskio low read performance X-BeenThere: freebsd-performance@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Performance/tuning List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 13 Jan 2007 20:03:20 -0000 Kris Kennaway disse na ultima mensagem: >> >> But also I checked the ULE/BSD against my particular problem and there >> is >> no difference at all. I get no acceptable disk read performance when >> comparing what I had with 4.11, wether with ULE or with 4BSD > > Be very careful, because I've personally measured severe disk I/O > penalties with ULE on SMP hardware. In fact in my testing ULE gives > worse SMP performance under load across the board compared to 4BSD > (it's only faster for the lightest of workloads). > seems suspicious and I will compile this night and recheck to be sure > If you're absolutely certain that ULE is not to blame (and want to > continue to take the risk of other performance and stability problems > down the line), that basically leaves something to do with the scsi > driver and/or its interaction with your hardware as the probable > cause. I don't know enough about this particular hardware to comment > further though. I will check 4BSD again, but I do not have stability problems. The system works really constant and without any odds but the diskio thing certainly the reason why I have a LSI card is that I was blaming the adaptec driver first, but even with the adaptec disabled and the LSI card on the Supermicro nothing changed. So I do not know for sure but unlikely both drivers have the same problem I guess. > > * adaptive mutexes are usually a win so it's a bit unusual that you > have disabled them, but I assume you have tested this. > I get better network performance when not using polling with it and seems not to harm polling in my case so I let it in > * Dunno about the AUTO_EOI_1 option, I don't think you even have this > hardware on your system (device atpic didn't probe in your dmesg). right, it is a old setting I ever forget taking it out > > * HZ=1000 is superfluous since it is the default but may or may not > help. In some workloads the increased overhead relative to the old > default of HZ=100 gives a performance loss. Maybe it helps with > polling though, I dunno. that I did not know but does no bad either. With lower HZ it does not work for me and I have the impression it works better with 2000. I have 5 NICs and that is the reason I guess. But it is not my priority at this moment and I believe it does not mess with diskio. Michel computador é como nem cavalo e mulher mais que montam neles, pior que ficam ... **************************************************** Datacenter Matik http://datacenter.matik.com.br E-Mail e Data Hosting Service para Profissionais. ****************************************************