From owner-freebsd-fs Sun Oct 4 00:02:06 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id AAA24752 for freebsd-fs-outgoing; Sun, 4 Oct 1998 00:02:06 -0700 (PDT) (envelope-from owner-freebsd-fs@FreeBSD.ORG) Received: from dingo.cdrom.com (castles144.castles.com [208.214.165.144]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id AAA24668; Sun, 4 Oct 1998 00:01:39 -0700 (PDT) (envelope-from mike@dingo.cdrom.com) Received: from dingo.cdrom.com (localhost [127.0.0.1]) by dingo.cdrom.com (8.9.1/8.8.8) with ESMTP id AAA02436; Sun, 4 Oct 1998 00:06:16 -0700 (PDT) (envelope-from mike@dingo.cdrom.com) Message-Id: <199810040706.AAA02436@dingo.cdrom.com> X-Mailer: exmh version 2.0.2 2/24/98 To: Bruce Evans cc: mike@smith.net.au, Don.Lewis@tsc.tdk.com, freebsd-fs@FreeBSD.ORG, freebsd-scsi@FreeBSD.ORG, gibbs@plutotech.com, tlambert@primenet.com Subject: Re: filesystem safety and SCSI disk write caching In-reply-to: Your message of "Sun, 04 Oct 1998 16:16:24 +1000." <199810040616.QAA26536@godzilla.zeta.org.au> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Sun, 04 Oct 1998 00:06:15 -0700 From: Mike Smith Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org > >> Yes, the default configuration may be much slower than mine. > > > >I can definitely back your basic point ('make world' is CPU bound) up. > >On a 4-way Xeon system with slow disks we were still able to get down > >around 40 minutes. > > Er, that shows that it is i/o bound on systems with so much CPU. I > got it down to 75 minutes on 1-way K6-233 with 1 IDE disk before it > was bloated by perl5 and transition to elf. Moving to an MFS only saved about 15% of the build time. My point was that a faster CPU let you go faster. If the build was I/O bound, it wouldn't. -- \\ Sometimes you're ahead, \\ Mike Smith \\ sometimes you're behind. \\ mike@smith.net.au \\ The race is long, and in the \\ msmith@freebsd.org \\ end it's only with yourself. \\ msmith@cdrom.com To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message From owner-freebsd-fs Sun Oct 4 03:33:32 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id DAA15518 for freebsd-fs-outgoing; Sun, 4 Oct 1998 03:33:32 -0700 (PDT) (envelope-from owner-freebsd-fs@FreeBSD.ORG) Received: from post.mail.demon.net (post-11.mail.demon.net [194.217.242.40]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id DAA15457; Sun, 4 Oct 1998 03:32:45 -0700 (PDT) (envelope-from dmlb@ragnet.demon.co.uk) Received: from [158.152.46.40] (helo=ragnet.demon.co.uk) by post.mail.demon.net with smtp (Exim 2.03 #1) id 0zPlSe-0003Yz-00; Sun, 4 Oct 1998 10:32:20 +0000 Received: from dmlb by ragnet.demon.co.uk with local (Exim 1.82 #1) id 0zPlSW-0006jk-00; Sun, 4 Oct 1998 11:32:12 +0100 Message-ID: X-Mailer: XFMail 1.3 [p0] on FreeBSD X-Priority: 3 (Normal) Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 8bit MIME-Version: 1.0 In-Reply-To: <199810040616.QAA26536@godzilla.zeta.org.au> Date: Sun, 04 Oct 1998 11:32:12 +0100 (BST) From: Duncan Barclay To: Bruce Evans Subject: Re: filesystem safety and SCSI disk write caching Cc: tlambert@primenet.com, gibbs@plutotech.com, freebsd-scsi@FreeBSD.ORG, freebsd-fs@FreeBSD.ORG, Don.Lewis@tsc.tdk.com, mike@smith.net.au Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org On 04-Oct-98 Bruce Evans wrote: >>> Yes, the default configuration may be much slower than mine. >> >>I can definitely back your basic point ('make world' is CPU bound) up. >>On a 4-way Xeon system with slow disks we were still able to get down >>around 40 minutes. > > Er, that shows that it is i/o bound on systems with so much CPU. I > got it down to 75 minutes on 1-way K6-233 with 1 IDE disk before it > was bloated by perl5 and transition to elf. > > Bruce I would agree, my AMD K6-233 did a make -j 4 in jst over an hour with src on IDE mounted async and obj on SCSI mounted async (this was upgrading from 2.2.6). I haven't tried it under -3.0 and softupdates yet. Duncan --- ________________________________________________________________________ Duncan Barclay | God smiles upon the little children, dmlb@ragnet.demon.co.uk | the alcoholics, and the permanently stoned. ________________________________________________________________________ To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message From owner-freebsd-fs Sun Oct 4 06:26:11 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id GAA08973 for freebsd-fs-outgoing; Sun, 4 Oct 1998 06:26:11 -0700 (PDT) (envelope-from owner-freebsd-fs@FreeBSD.ORG) Received: from ns1.yes.no (ns1.yes.no [195.204.136.10]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id GAA08953; Sun, 4 Oct 1998 06:26:01 -0700 (PDT) (envelope-from eivind@bitbox.follo.net) Received: from bitbox.follo.net (bitbox.follo.net [195.204.143.218]) by ns1.yes.no (8.9.1a/8.9.1) with ESMTP id PAA07587; Sun, 4 Oct 1998 15:25:32 +0200 (CEST) Received: (from eivind@localhost) by bitbox.follo.net (8.8.8/8.8.6) id PAA00898; Sun, 4 Oct 1998 15:25:29 +0200 (MET DST) Message-ID: <19981004152528.12252@follo.net> Date: Sun, 4 Oct 1998 15:25:28 +0200 From: Eivind Eklund To: Mike Smith , Bruce Evans Cc: Don.Lewis@tsc.tdk.com, freebsd-fs@FreeBSD.ORG, freebsd-scsi@FreeBSD.ORG, gibbs@plutotech.com, tlambert@primenet.com Subject: Re: filesystem safety and SCSI disk write caching References: <199810040616.QAA26536@godzilla.zeta.org.au> <199810040706.AAA02436@dingo.cdrom.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.89.1i In-Reply-To: <199810040706.AAA02436@dingo.cdrom.com>; from Mike Smith on Sun, Oct 04, 1998 at 12:06:15AM -0700 Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org On Sun, Oct 04, 1998 at 12:06:15AM -0700, Mike Smith wrote: > > >> Yes, the default configuration may be much slower than mine. > > > > > >I can definitely back your basic point ('make world' is CPU bound) up. > > >On a 4-way Xeon system with slow disks we were still able to get down > > >around 40 minutes. > > > > Er, that shows that it is i/o bound on systems with so much CPU. I > > got it down to 75 minutes on 1-way K6-233 with 1 IDE disk before it > > was bloated by perl5 and transition to elf. > > Moving to an MFS only saved about 15% of the build time. My point was > that a faster CPU let you go faster. If the build was I/O bound, it > wouldn't. My hypothesis is that for the high end boxes, 'make world' is mostly bound by memory bandwidth. This is what seems to best match the speed patterns people have been reporting. Unfortunately, I can only think of a single way of verifying this: Running a PPro build in 64-bit mode and 128-bit mode, with the same amount of the same speed memory. It would require a motherboard that had 128-bit support; AFAIK, this is only present in the early Intel boards :-( Eivind. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message From owner-freebsd-fs Sun Oct 4 12:47:52 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id MAA16747 for freebsd-fs-outgoing; Sun, 4 Oct 1998 12:47:52 -0700 (PDT) (envelope-from owner-freebsd-fs@FreeBSD.ORG) Received: from the.nu (s232-128i32.9na.com [208.232.128.32]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id MAA16727 for ; Sun, 4 Oct 1998 12:47:45 -0700 (PDT) (envelope-from bsd@the.nu) Received: from localhost (bsd@localhost) by the.nu (8.8.7/8.8.8) with SMTP id PAA14630 for ; Sun, 4 Oct 1998 15:47:20 -0400 (EDT) Date: Sun, 4 Oct 1998 15:47:20 -0400 (EDT) From: "Mr. K." To: freebsd-fs@FreeBSD.ORG Subject: efficient filesystem Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org does freebsd have a filesystem which would be efficient with 10,000+ files in a single directory? I have other possibilites such as manually breaking the files into directories (f/i/file.gif) or possibly using my mysql database but a filesystem which could add/delete/find files in log n time is what I'm really looking for. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message From owner-freebsd-fs Sun Oct 4 13:24:39 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id NAA21159 for freebsd-fs-outgoing; Sun, 4 Oct 1998 13:24:39 -0700 (PDT) (envelope-from owner-freebsd-fs@FreeBSD.ORG) Received: from gjp.erols.com (alex-va-n008c079.moon.jic.com [206.156.18.89]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id NAA21154 for ; Sun, 4 Oct 1998 13:24:37 -0700 (PDT) (envelope-from gjp@gjp.erols.com) Received: from gjp.erols.com (gjp@localhost.erols.com [127.0.0.1]) by gjp.erols.com (8.8.8/8.8.7) with ESMTP id QAA17829; Sun, 4 Oct 1998 16:23:30 -0400 (EDT) (envelope-from gjp@gjp.erols.com) X-Mailer: exmh version 2.0.1 12/23/97 To: "Mr. K." cc: freebsd-fs@FreeBSD.ORG From: "Gary Palmer" Subject: Re: efficient filesystem In-reply-to: Your message of "Sun, 04 Oct 1998 15:47:20 EDT." Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Sun, 04 Oct 1998 16:23:30 -0400 Message-ID: <17825.907532610@gjp.erols.com> Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org "Mr. K." wrote in message ID : > does freebsd have a filesystem which would be efficient with 10,000+ files > in a single directory? I have other possibilites such as manually > breaking the files into directories (f/i/file.gif) or possibly using my > mysql database but a filesystem which could add/delete/find files in log n > time is what I'm really looking for. Do the directory breakout shuffle. Without something like VxVS (Veritas Filesystem, not available for non-commercial OSen unfortunately), which has some sort of hashed directory structure if I understand it right, then you are reduced to linear searches for all operations. Even with VxFS, maintaining a large directory like that becomes unweidly for humans, and when I was running a 300,000 user mail system, we still used /var/mail///username/Mailbox .... Breaking it out into subdirectories makes everyones lives easier, including the machines .... Gary -- Gary Palmer FreeBSD Core Team Member FreeBSD: Turning PC's into workstations. See http://www.FreeBSD.ORG/ for info To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message From owner-freebsd-fs Sun Oct 4 17:02:24 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id RAA21856 for freebsd-fs-outgoing; Sun, 4 Oct 1998 17:02:24 -0700 (PDT) (envelope-from owner-freebsd-fs@FreeBSD.ORG) Received: from smtp02.primenet.com (smtp02.primenet.com [206.165.6.132]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id RAA21851; Sun, 4 Oct 1998 17:02:15 -0700 (PDT) (envelope-from tlambert@usr09.primenet.com) Received: (from daemon@localhost) by smtp02.primenet.com (8.8.8/8.8.8) id RAA03786; Sun, 4 Oct 1998 17:01:53 -0700 (MST) Received: from usr09.primenet.com(206.165.6.209) via SMTP by smtp02.primenet.com, id smtpd003739; Sun Oct 4 17:01:46 1998 Received: (from tlambert@localhost) by usr09.primenet.com (8.8.5/8.8.5) id RAA09338; Sun, 4 Oct 1998 17:01:42 -0700 (MST) From: Terry Lambert Message-Id: <199810050001.RAA09338@usr09.primenet.com> Subject: Re: efficient filesystem To: gpalmer@FreeBSD.ORG (Gary Palmer) Date: Mon, 5 Oct 1998 00:01:41 +0000 (GMT) Cc: bsd@the.nu, freebsd-fs@FreeBSD.ORG In-Reply-To: <17825.907532610@gjp.erols.com> from "Gary Palmer" at Oct 4, 98 04:23:30 pm X-Mailer: ELM [version 2.4 PL25] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org > Do the directory breakout shuffle. Without something like VxVS (Veritas > Filesystem, not available for non-commercial OSen unfortunately), which has > some sort of hashed directory structure if I understand it right, then you are Actually, VxFS's directory code is lifted, pretty much verbatim, from the AT&T UFS (FFS) implementation. It has AT&T Copyrights all over the sources. What you need, if you are going to do something so silly as to have a huge flat directory structure, is to implement a btree structured directory, like OS/2 HPFS. My advice is to not do this; instead you should break it down. The number of entries that have to be traversed to look up a directory entry is 512 div (average_file_name_length + 24) * number of blocks. Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message From owner-freebsd-fs Sun Oct 4 17:40:36 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id RAA26282 for freebsd-fs-outgoing; Sun, 4 Oct 1998 17:40:36 -0700 (PDT) (envelope-from owner-freebsd-fs@FreeBSD.ORG) Received: from gjp.erols.com (alex-va-n008c079.moon.jic.com [206.156.18.89]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id RAA26264 for ; Sun, 4 Oct 1998 17:40:26 -0700 (PDT) (envelope-from gjp@gjp.erols.com) Received: from gjp.erols.com (gjp@localhost.erols.com [127.0.0.1]) by gjp.erols.com (8.8.8/8.8.7) with ESMTP id UAA21415; Sun, 4 Oct 1998 20:39:42 -0400 (EDT) (envelope-from gjp@gjp.erols.com) X-Mailer: exmh version 2.0.1 12/23/97 To: Terry Lambert cc: bsd@the.nu, freebsd-fs@FreeBSD.ORG From: "Gary Palmer" Subject: Re: efficient filesystem In-reply-to: Your message of "Mon, 05 Oct 1998 00:01:41 -0000." <199810050001.RAA09338@usr09.primenet.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Sun, 04 Oct 1998 20:39:42 -0400 Message-ID: <21411.907547982@gjp.erols.com> Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org Terry Lambert wrote in message ID <199810050001.RAA09338@usr09.primenet.com>: > Actually, VxFS's directory code is lifted, pretty much verbatim, from > the AT&T UFS (FFS) implementation. It has AT&T Copyrights all over > the sources. Which version of VxFS did you look at? Because I'm pretty certain that I was told by someone who would know that that is no longer the case, and that it does non-linear searching. Gary -- Gary Palmer FreeBSD Core Team Member FreeBSD: Turning PC's into workstations. See http://www.FreeBSD.ORG/ for info To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message From owner-freebsd-fs Sun Oct 4 17:50:35 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id RAA27744 for freebsd-fs-outgoing; Sun, 4 Oct 1998 17:50:35 -0700 (PDT) (envelope-from owner-freebsd-fs@FreeBSD.ORG) Received: from the.nu (s232-128i32.9na.com [208.232.128.32]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id RAA27714 for ; Sun, 4 Oct 1998 17:50:20 -0700 (PDT) (envelope-from bsd@the.nu) Received: from localhost (bsd@localhost) by the.nu (8.8.7/8.8.8) with SMTP id UAA21002 for ; Sun, 4 Oct 1998 20:50:03 -0400 (EDT) Date: Sun, 4 Oct 1998 20:50:03 -0400 (EDT) From: "Mr. K." To: freebsd-fs@FreeBSD.ORG Subject: Re: efficient filesystem (fwd) Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org On Mon, 5 Oct 1998, Terry Lambert wrote: > > Do the directory breakout shuffle. Without something like VxVS (Veritas > > Filesystem, not available for non-commercial OSen unfortunately), which has > > some sort of hashed directory structure if I understand it right, then you are > > Actually, VxFS's directory code is lifted, pretty much verbatim, from > the AT&T UFS (FFS) implementation. It has AT&T Copyrights all over > the sources. > perhaps this is an old version you were looking at? http://compassion.qualix.com/sysman/product/vxvm.htmld/vsm-over.html "VxFS supports variable-length directory entries. Because file names may be very long, rather than allocate a directory entry at maximum size, wasting directory space and slowing access time, variable-length entries are used. However, these are less efficient to search than fixed-sized entries, because you cannot know where the next entry begins. To eliminate the need for a linear directory search, VxFS uses a hash table in each directory extent to locate directory entries quickly, accelerating file access." > What you need, if you are going to do something so silly as to have > a huge flat directory structure, is to implement a btree structured > directory, like OS/2 HPFS. > yes, I agree, and I was wondering if anyone had implemented something like this for freebsd. It's not worth it for me to implement it myself, and I'm not about to waste money on something commercial. Then again, if you're hacking on top of u/h/ffs, the hashing strategy is probably easier. I've also seen it done completely in the dnlc, which would keep complete compatibility, but it takes up memory and you have to load in all the data the first time you search each directory. > My advice is to not do this; instead you should break it down. > for now that's what I'll do but it's going to kludge up my cgi program and I'll have to scale it manually as I grow, so eventually I'll get to putting my files into a mysql database and have the webserver get the files from there. > The number of entries that have to be traversed to look up a directory > entry is 512 div (average_file_name_length + 24) * number of blocks. > > > Terry Lambert > terry@lambert.org > --- > Any opinions in this posting are my own and not those of my present > or previous employers. > To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message From owner-freebsd-fs Sun Oct 4 18:08:04 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id SAA00229 for freebsd-fs-outgoing; Sun, 4 Oct 1998 18:08:04 -0700 (PDT) (envelope-from owner-freebsd-fs@FreeBSD.ORG) Received: from slarti.muc.de (slarti.muc.de [193.174.4.10]) by hub.freebsd.org (8.8.8/8.8.8) with SMTP id SAA00193 for ; Sun, 4 Oct 1998 18:07:51 -0700 (PDT) (envelope-from eilts@tor.muc.de) Received: (qmail 5712 invoked by uid 66); 5 Oct 1998 01:07:12 -0000 Received: (from eilts@localhost) by tor.muc.de (8.8.7/8.6.6) id QAA02453; Sun, 4 Oct 1998 16:00:01 +0200 (CEST) Date: Sun, 4 Oct 1998 16:00:01 +0200 (CEST) From: Hinrich Eilts Message-Id: <199810041400.QAA02453@tor.muc.de> To: freebsd-fs@FreeBSD.ORG Subject: Re: filesystem safety and SCSI disk write caching X-Newsreader: NN version 6.5.0 CURRENT #11 Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org > I can post (once again) the results of a Novell study on server usage > patterns. The 30,000 foot view for a typical server breaks down to: > > 75% reads > 15% writes > 8% directory search operations > 2% other How does this translate to disk usage? I think, VM/buffer cache will reduce reads and directory search operations quite a lot. Hinrich To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message From owner-freebsd-fs Sun Oct 4 18:55:43 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id SAA06382 for freebsd-fs-outgoing; Sun, 4 Oct 1998 18:55:43 -0700 (PDT) (envelope-from owner-freebsd-fs@FreeBSD.ORG) Received: from sv01.cet.co.jp (sv01.cet.co.jp [210.171.56.2]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id SAA06363 for ; Sun, 4 Oct 1998 18:55:37 -0700 (PDT) (envelope-from michaelh@cet.co.jp) Received: from localhost (michaelh@localhost) by sv01.cet.co.jp (8.8.8/8.8.8) with SMTP id BAA02500; Mon, 5 Oct 1998 01:48:11 GMT (envelope-from michaelh@cet.co.jp) Date: Mon, 5 Oct 1998 10:48:11 +0900 (JST) From: Michael Hancock To: Hinrich Eilts cc: freebsd-fs@FreeBSD.ORG Subject: Re: filesystem safety and SCSI disk write caching In-Reply-To: <199810041400.QAA02453@tor.muc.de> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org On Sun, 4 Oct 1998, Hinrich Eilts wrote: > > I can post (once again) the results of a Novell study on server usage > > patterns. The 30,000 foot view for a typical server breaks down to: > > > > 75% reads > > 15% writes > > 8% directory search operations > > 2% other > > How does this translate to disk usage? I think, VM/buffer cache will > reduce reads and directory search operations quite a lot. No. Yes. The ratio of actual disk writes to disk reads is probably enough to justify the effort to optimize writes. Regards, Mike Hancock To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message From owner-freebsd-fs Sun Oct 4 19:08:09 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id TAA07768 for freebsd-fs-outgoing; Sun, 4 Oct 1998 19:08:09 -0700 (PDT) (envelope-from owner-freebsd-fs@FreeBSD.ORG) Received: from gjp.erols.com (alex-va-n008c079.moon.jic.com [206.156.18.89]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id TAA07711 for ; Sun, 4 Oct 1998 19:08:04 -0700 (PDT) (envelope-from gjp@gjp.erols.com) Received: from gjp.erols.com (gjp@localhost.erols.com [127.0.0.1]) by gjp.erols.com (8.8.8/8.8.7) with ESMTP id WAA22661 for ; Sun, 4 Oct 1998 22:07:42 -0400 (EDT) (envelope-from gjp@gjp.erols.com) X-Mailer: exmh version 2.0.1 12/23/97 To: freebsd-fs@FreeBSD.ORG From: "Gary Palmer" Subject: Re: filesystem safety and SCSI disk write caching In-reply-to: Your message of "Sun, 04 Oct 1998 16:00:01 +0200." <199810041400.QAA02453@tor.muc.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Sun, 04 Oct 1998 22:07:42 -0400 Message-ID: <22657.907553262@gjp.erols.com> Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org > > I can post (once again) the results of a Novell study on server usage > > patterns. The 30,000 foot view for a typical server breaks down to: > > > > 75% reads > > 15% writes > > 8% directory search operations > > 2% other I think that is very dependant on the server type. PC NetWare fileservers probably have very different access patterns to (say) a web server or a mail server. Let alone a news server. Gary -- Gary Palmer FreeBSD Core Team Member FreeBSD: Turning PC's into workstations. See http://www.FreeBSD.ORG/ for info To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message From owner-freebsd-fs Sun Oct 4 20:59:03 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id UAA24299 for freebsd-fs-outgoing; Sun, 4 Oct 1998 20:59:03 -0700 (PDT) (envelope-from owner-freebsd-fs@FreeBSD.ORG) Received: from pluto.plutotech.com (mail.plutotech.com [206.168.67.137]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id UAA24146; Sun, 4 Oct 1998 20:58:24 -0700 (PDT) (envelope-from gibbs@plutotech.com) Received: from narnia.plutotech.com (narnia.plutotech.com [206.168.67.130]) by pluto.plutotech.com (8.8.7/8.8.5) with ESMTP id VAA05121; Sun, 4 Oct 1998 21:57:52 -0600 (MDT) Message-Id: <199810050357.VAA05121@pluto.plutotech.com> X-Mailer: exmh version 2.0.2 2/24/98 To: Terry Lambert cc: gibbs@plutotech.com (Justin T. Gibbs), julian@whistle.com, Don.Lewis@tsc.tdk.com, freebsd-fs@FreeBSD.ORG, freebsd-scsi@FreeBSD.ORG Subject: Re: filesystem safety and SCSI disk write caching In-reply-to: Your message of "Sat, 03 Oct 1998 01:08:48 -0000." <199810030108.SAA02581@usr06.primenet.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Sun, 04 Oct 1998 21:51:23 -0600 From: "Justin T. Gibbs" Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org >> >I told you so. >> >> You told me some things that were in-correct and some things that >> I already knew. Par for the course. > >Feel free to make his setup work with SCSI write caching enabled. I gave the recipe for this on freebsd-alpha near the end of september. 1) Use a UPS. 2) Use a drive with non-bogus firmware. Recent Seagate and IBM drives should work just fine. I haven't validated any Quantum drives in this regard yet. -- Justin To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message From owner-freebsd-fs Mon Oct 5 07:39:49 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id HAA15987 for freebsd-fs-outgoing; Mon, 5 Oct 1998 07:39:49 -0700 (PDT) (envelope-from owner-freebsd-fs@FreeBSD.ORG) Received: from opi.flirtbox.ch ([62.48.0.50]) by hub.freebsd.org (8.8.8/8.8.8) with SMTP id HAA15969 for ; Mon, 5 Oct 1998 07:39:44 -0700 (PDT) (envelope-from oppermann@pipeline.ch) Received: (qmail 11966 invoked from network); 5 Oct 1998 14:39:56 -0000 Received: from unknown (HELO pipeline.ch) (195.134.140.3) by opi.flirtbox.ch with SMTP; 5 Oct 1998 14:39:56 -0000 Message-ID: <3618DA15.46329AA1@pipeline.ch> Date: Mon, 05 Oct 1998 16:39:17 +0200 From: Andre Oppermann Organization: Internet Business Solutions Ltd. X-Mailer: Mozilla 4.03 [en] (Win95; I) MIME-Version: 1.0 To: Gary Palmer CC: freebsd-fs@FreeBSD.ORG Subject: Re: filesystem safety and SCSI disk write caching References: <22657.907553262@gjp.erols.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org Gary Palmer wrote: > > > > I can post (once again) the results of a Novell study on server usage > > > patterns. The 30,000 foot view for a typical server breaks down to: > > > > > > 75% reads > > > 15% writes > > > 8% directory search operations > > > 2% other > > I think that is very dependant on the server type. PC NetWare fileservers > probably have very different access patterns to (say) a web server or a mail > server. Let alone a news server. Is there a way to gather such statistics on FreeBSD? I'd like to run it on all my boxes (and others) to get representative figures. After that we can discuss optimizations. -- Andre To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message From owner-freebsd-fs Mon Oct 5 10:31:56 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id KAA17965 for freebsd-fs-outgoing; Mon, 5 Oct 1998 10:31:56 -0700 (PDT) (envelope-from owner-freebsd-fs@FreeBSD.ORG) Received: from smtp04.primenet.com (smtp04.primenet.com [206.165.6.134]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id KAA17935 for ; Mon, 5 Oct 1998 10:31:47 -0700 (PDT) (envelope-from tlambert@usr01.primenet.com) Received: (from daemon@localhost) by smtp04.primenet.com (8.8.8/8.8.8) id KAA05860; Mon, 5 Oct 1998 10:31:24 -0700 (MST) Received: from usr01.primenet.com(206.165.6.201) via SMTP by smtp04.primenet.com, id smtpd005708; Mon Oct 5 10:31:17 1998 Received: (from tlambert@localhost) by usr01.primenet.com (8.8.5/8.8.5) id KAA19781; Mon, 5 Oct 1998 10:31:05 -0700 (MST) From: Terry Lambert Message-Id: <199810051731.KAA19781@usr01.primenet.com> Subject: Re: filesystem safety and SCSI disk write caching To: eilts@tor.muc.de (Hinrich Eilts) Date: Mon, 5 Oct 1998 17:31:05 +0000 (GMT) Cc: freebsd-fs@FreeBSD.ORG In-Reply-To: <199810041400.QAA02453@tor.muc.de> from "Hinrich Eilts" at Oct 4, 98 04:00:01 pm X-Mailer: ELM [version 2.4 PL25] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org > > I can post (once again) the results of a Novell study on server usage > > patterns. The 30,000 foot view for a typical server breaks down to: > > > > 75% reads > > 15% writes > > 8% directory search operations > > 2% other > > How does this translate to disk usage? I think, VM/buffer cache will > reduce reads and directory search operations quite a lot. I'm only reporting what Novell has characterized as "typical server usage" over a much larger statistical smaple than has ever been used in the UNIX community. You're free to draw your own conclusions about what "typical server usage" means. My conclusion is that reads are 5 times more frequent than writes and 10 times more frequent than directory operations. Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message From owner-freebsd-fs Mon Oct 5 10:50:55 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id KAA20573 for freebsd-fs-outgoing; Mon, 5 Oct 1998 10:50:55 -0700 (PDT) (envelope-from owner-freebsd-fs@FreeBSD.ORG) Received: from smtp01.primenet.com (smtp01.primenet.com [206.165.6.131]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id KAA20539; Mon, 5 Oct 1998 10:50:45 -0700 (PDT) (envelope-from tlambert@usr01.primenet.com) Received: (from daemon@localhost) by smtp01.primenet.com (8.8.8/8.8.8) id KAA25206; Mon, 5 Oct 1998 10:50:18 -0700 (MST) Received: from usr01.primenet.com(206.165.6.201) via SMTP by smtp01.primenet.com, id smtpd025174; Mon Oct 5 10:50:13 1998 Received: (from tlambert@localhost) by usr01.primenet.com (8.8.5/8.8.5) id KAA20600; Mon, 5 Oct 1998 10:50:09 -0700 (MST) From: Terry Lambert Message-Id: <199810051750.KAA20600@usr01.primenet.com> Subject: Re: efficient filesystem To: gpalmer@FreeBSD.ORG (Gary Palmer) Date: Mon, 5 Oct 1998 17:50:09 +0000 (GMT) Cc: tlambert@primenet.com, bsd@the.nu, freebsd-fs@FreeBSD.ORG In-Reply-To: <21411.907547982@gjp.erols.com> from "Gary Palmer" at Oct 4, 98 08:39:42 pm X-Mailer: ELM [version 2.4 PL25] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org > > Actually, VxFS's directory code is lifted, pretty much verbatim, from > > the AT&T UFS (FFS) implementation. It has AT&T Copyrights all over > > the sources. > > Which version of VxFS did you look at? Because I'm pretty certain that I was > told by someone who would know that that is no longer the case, and that it > does non-linear searching. I hacked on the one USL was working on 4 years ago. After a little prodding, and a little looking, it looks like they are using a hash table "I've worked with a btree, and hash table, you're no btree". It would be rather trivial to make something like this work in an existing FFS, taking advantage of the use of inode 0 to indicate an empty directory entry and that inode 1 is not used because it was historically used for bad blocks. Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message From owner-freebsd-fs Mon Oct 5 10:57:47 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id KAA21886 for freebsd-fs-outgoing; Mon, 5 Oct 1998 10:57:47 -0700 (PDT) (envelope-from owner-freebsd-fs@FreeBSD.ORG) Received: from smtp01.primenet.com (smtp01.primenet.com [206.165.6.131]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id KAA21879; Mon, 5 Oct 1998 10:57:42 -0700 (PDT) (envelope-from tlambert@usr01.primenet.com) Received: (from daemon@localhost) by smtp01.primenet.com (8.8.8/8.8.8) id KAA28142; Mon, 5 Oct 1998 10:57:17 -0700 (MST) Received: from usr01.primenet.com(206.165.6.201) via SMTP by smtp01.primenet.com, id smtpd028058; Mon Oct 5 10:57:08 1998 Received: (from tlambert@localhost) by usr01.primenet.com (8.8.5/8.8.5) id KAA20911; Mon, 5 Oct 1998 10:57:02 -0700 (MST) From: Terry Lambert Message-Id: <199810051757.KAA20911@usr01.primenet.com> Subject: Re: filesystem safety and SCSI disk write caching To: gibbs@plutotech.com (Justin T. Gibbs) Date: Mon, 5 Oct 1998 17:57:01 +0000 (GMT) Cc: tlambert@primenet.com, gibbs@plutotech.com, julian@whistle.com, Don.Lewis@tsc.tdk.com, freebsd-fs@FreeBSD.ORG, freebsd-scsi@FreeBSD.ORG In-Reply-To: <199810050357.VAA05121@pluto.plutotech.com> from "Justin T. Gibbs" at Oct 4, 98 09:51:23 pm X-Mailer: ELM [version 2.4 PL25] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org > I gave the recipe for this on freebsd-alpha near the end of september. > > 1) Use a UPS. > > 2) Use a drive with non-bogus firmware. Recent Seagate and IBM > drives should work just fine. I haven't validated any Quantum > drives in this regard yet. I think the base assumption should be that the firmware is bogus, unless proven otherwise. You could have a "known good" table; I think a "known rogues" table would be too large. 8-(. Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message From owner-freebsd-fs Mon Oct 5 11:02:53 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id LAA23239 for freebsd-fs-outgoing; Mon, 5 Oct 1998 11:02:53 -0700 (PDT) (envelope-from owner-freebsd-fs@FreeBSD.ORG) Received: from smtp01.primenet.com (smtp01.primenet.com [206.165.6.131]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id LAA23223 for ; Mon, 5 Oct 1998 11:02:48 -0700 (PDT) (envelope-from tlambert@usr01.primenet.com) Received: (from daemon@localhost) by smtp01.primenet.com (8.8.8/8.8.8) id LAA00353; Mon, 5 Oct 1998 11:02:19 -0700 (MST) Received: from usr01.primenet.com(206.165.6.201) via SMTP by smtp01.primenet.com, id smtpd000240; Mon Oct 5 11:02:06 1998 Received: (from tlambert@localhost) by usr01.primenet.com (8.8.5/8.8.5) id LAA21073; Mon, 5 Oct 1998 11:01:59 -0700 (MST) From: Terry Lambert Message-Id: <199810051801.LAA21073@usr01.primenet.com> Subject: Re: filesystem safety and SCSI disk write caching To: michaelh@cet.co.jp (Michael Hancock) Date: Mon, 5 Oct 1998 18:01:59 +0000 (GMT) Cc: eilts@tor.muc.de, freebsd-fs@FreeBSD.ORG In-Reply-To: from "Michael Hancock" at Oct 5, 98 10:48:11 am X-Mailer: ELM [version 2.4 PL25] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org > > > I can post (once again) the results of a Novell study on server usage > > > patterns. The 30,000 foot view for a typical server breaks down to: > > > > > > 75% reads > > > 15% writes > > > 8% directory search operations > > > 2% other > > > > How does this translate to disk usage? I think, VM/buffer cache will > > reduce reads and directory search operations quite a lot. > > No. Yes. The ratio of actual disk writes to disk reads is probably enough > to justify the effort to optimize writes. According to other postings in this thread, the empirical difference in wall time measured on the prototypical FreeBSD benchmark of "make world" is 5.6% between async + noatime vs. soft updates. While it would be worthwhile to attempt to optimize this further, I think doing so by enabling write caching is a cop-out. Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message From owner-freebsd-fs Mon Oct 5 11:07:34 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id LAA24406 for freebsd-fs-outgoing; Mon, 5 Oct 1998 11:07:34 -0700 (PDT) (envelope-from owner-freebsd-fs@FreeBSD.ORG) Received: from smtp01.primenet.com (smtp01.primenet.com [206.165.6.131]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id LAA24381; Mon, 5 Oct 1998 11:07:24 -0700 (PDT) (envelope-from tlambert@usr01.primenet.com) Received: (from daemon@localhost) by smtp01.primenet.com (8.8.8/8.8.8) id LAA02027; Mon, 5 Oct 1998 11:06:14 -0700 (MST) Received: from usr01.primenet.com(206.165.6.201) via SMTP by smtp01.primenet.com, id smtpd001974; Mon Oct 5 11:06:05 1998 Received: (from tlambert@localhost) by usr01.primenet.com (8.8.5/8.8.5) id LAA21513; Mon, 5 Oct 1998 11:06:04 -0700 (MST) From: Terry Lambert Message-Id: <199810051806.LAA21513@usr01.primenet.com> Subject: Re: filesystem safety and SCSI disk write caching To: gpalmer@FreeBSD.ORG (Gary Palmer) Date: Mon, 5 Oct 1998 18:06:04 +0000 (GMT) Cc: freebsd-fs@FreeBSD.ORG In-Reply-To: <22657.907553262@gjp.erols.com> from "Gary Palmer" at Oct 4, 98 10:07:42 pm X-Mailer: ELM [version 2.4 PL25] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org > > > I can post (once again) the results of a Novell study on server usage > > > patterns. The 30,000 foot view for a typical server breaks down to: > > > > > > 75% reads > > > 15% writes > > > 8% directory search operations > > > 2% other > > I think that is very dependant on the server type. PC NetWare fileservers > probably have very different access patterns to (say) a web server or a mail > server. Let alone a news server. That depends. News servers are "read mostly". HTTP servers are "read mostly". IMAP4 servers, if they are used as intended, are "read mostly". POP3 servers are not. SMTP servers are not. SMB servers and AppleTalk servers probably show exactly the access patterns Novell measuered for NCP servers, since the requests are client-driven, as are those to NFS servers. But yes, this list is probably the optimization weighting for a file server as opposed to an application server. I've always disliked the term "application server", which was coined by Novell when it decided to not compete against Microsoft on the destop. 8-(. Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message From owner-freebsd-fs Mon Oct 5 11:51:09 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id LAA01844 for freebsd-fs-outgoing; Mon, 5 Oct 1998 11:51:09 -0700 (PDT) (envelope-from owner-freebsd-fs@FreeBSD.ORG) Received: from smtp02.primenet.com (smtp02.primenet.com [206.165.6.132]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id LAA01779; Mon, 5 Oct 1998 11:50:47 -0700 (PDT) (envelope-from tlambert@usr01.primenet.com) Received: (from daemon@localhost) by smtp02.primenet.com (8.8.8/8.8.8) id LAA17982; Mon, 5 Oct 1998 11:50:14 -0700 (MST) Received: from usr01.primenet.com(206.165.6.201) via SMTP by smtp02.primenet.com, id smtpd017960; Mon Oct 5 11:50:08 1998 Received: (from tlambert@localhost) by usr01.primenet.com (8.8.5/8.8.5) id LAA23549; Mon, 5 Oct 1998 11:50:05 -0700 (MST) From: Terry Lambert Message-Id: <199810051850.LAA23549@usr01.primenet.com> Subject: Re: filesystem safety and SCSI disk write caching To: oppermann@pipeline.ch (Andre Oppermann) Date: Mon, 5 Oct 1998 18:50:04 +0000 (GMT) Cc: gpalmer@FreeBSD.ORG, freebsd-fs@FreeBSD.ORG In-Reply-To: <3618DA15.46329AA1@pipeline.ch> from "Andre Oppermann" at Oct 5, 98 04:39:17 pm X-Mailer: ELM [version 2.4 PL25] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org > > > > I can post (once again) the results of a Novell study on server usage > > > > patterns. The 30,000 foot view for a typical server breaks down to: > > > > > > > > 75% reads > > > > 15% writes > > > > 8% directory search operations > > > > 2% other > > > > I think that is very dependant on the server type. PC NetWare fileservers > > probably have very different access patterns to (say) a web server or a mail > > server. Let alone a news server. > > Is there a way to gather such statistics on FreeBSD? > > I'd like to run it on all my boxes (and others) to get representative > figures. After that we can discuss optimizations. This would be a very good idea. It would be best to instrument at the lowest level, *below* where soft updates does its implicit write gathering, and/or any elevator sorting occurs. I have suggested one set of instrumentation, which is interval sampling of number of tagged commands outstanding, which would resolve once and for all the current debate. Unfortunately, I don't have the good firmware that Justin seems to have, and thus I can't really make the measurements myself. It also seemed to me that the authors of the CAM code would be in a better position about where to validly determine max tag loading than I, and so would be better suited to the task of coding up the statistics gathering. One method of testing FreeBSD, but which would include testing SAMBA, would be to run the Ziff Davis "NetBench" suite against a FreeBSD machine acting as a server. There are similar benchmarks (LANBench?) available for download from Novell, laast time I looked at the FTP site. Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message From owner-freebsd-fs Mon Oct 5 12:03:09 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id MAA04934 for freebsd-fs-outgoing; Mon, 5 Oct 1998 12:03:09 -0700 (PDT) (envelope-from owner-freebsd-fs@FreeBSD.ORG) Received: from gjp.erols.com (alex-va-n008c079.moon.jic.com [206.156.18.89]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id MAA04916 for ; Mon, 5 Oct 1998 12:03:03 -0700 (PDT) (envelope-from gjp@gjp.erols.com) Received: from gjp.erols.com (gjp@localhost.erols.com [127.0.0.1]) by gjp.erols.com (8.8.8/8.8.7) with ESMTP id PAA05893; Mon, 5 Oct 1998 15:02:31 -0400 (EDT) (envelope-from gjp@gjp.erols.com) X-Mailer: exmh version 2.0.1 12/23/97 To: Terry Lambert cc: oppermann@pipeline.ch (Andre Oppermann), freebsd-fs@FreeBSD.ORG Reply-To: freebsd-fs@FreeBSD.ORG From: "Gary Palmer" Subject: Re: filesystem safety and SCSI disk write caching In-reply-to: Your message of "Mon, 05 Oct 1998 18:50:04 -0000." <199810051850.LAA23549@usr01.primenet.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Mon, 05 Oct 1998 15:02:31 -0400 Message-ID: <5889.907614151@gjp.erols.com> Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org Terry Lambert wrote in message ID <199810051850.LAA23549@usr01.primenet.com>: > > > > > I can post (once again) the results of a Novell study on server usage > > > > > patterns. The 30,000 foot view for a typical server breaks down to: > > > > > > > > > > 75% reads > > > > > 15% writes > > > > > 8% directory search operations > > > > > 2% other > This would be a very good idea. > > It would be best to instrument at the lowest level, *below* where > soft updates does its implicit write gathering, and/or any elevator > sorting occurs. Depends on what you want to gather. The stats quotes above would seem to be (IMHO) application<->FS layer transactions, not FS<->disk transactions. And to be honest, you'd need both sets of figures (as well as a good bit of analysis code which goes far beyond percentages) to obtain any sort of `optimizations' based off those figures. It would be really interesting to instrument at the application<->FS layer a few of the different server types (POP3/MTA only, IMAP/MTA only, news, etc) and see just what the access patterns are. I'm personally very curious about the POP3/MTA systems using off-the-shelf software. Gary -- Gary Palmer FreeBSD Core Team Member FreeBSD: Turning PC's into workstations. See http://www.FreeBSD.ORG/ for info To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message From owner-freebsd-fs Mon Oct 5 12:04:17 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id MAA05189 for freebsd-fs-outgoing; Mon, 5 Oct 1998 12:04:17 -0700 (PDT) (envelope-from owner-freebsd-fs@FreeBSD.ORG) Received: from gjp.erols.com (alex-va-n008c079.moon.jic.com [206.156.18.89]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id MAA05087 for ; Mon, 5 Oct 1998 12:03:57 -0700 (PDT) (envelope-from gjp@gjp.erols.com) Received: from gjp.erols.com (gjp@localhost.erols.com [127.0.0.1]) by gjp.erols.com (8.8.8/8.8.7) with ESMTP id PAA05917; Mon, 5 Oct 1998 15:03:32 -0400 (EDT) (envelope-from gjp@gjp.erols.com) X-Mailer: exmh version 2.0.1 12/23/97 To: Andre Oppermann cc: freebsd-fs@FreeBSD.ORG Reply-To: FreeBSD-FS@FreeBSD.ORG From: "Gary Palmer" Subject: Re: filesystem safety and SCSI disk write caching In-reply-to: Your message of "Mon, 05 Oct 1998 16:39:17 +0200." <3618DA15.46329AA1@pipeline.ch> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Mon, 05 Oct 1998 15:03:32 -0400 Message-ID: <5914.907614212@gjp.erols.com> Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org Andre Oppermann wrote in message ID <3618DA15.46329AA1@pipeline.ch>: > Is there a way to gather such statistics on FreeBSD? > > I'd like to run it on all my boxes (and others) to get representative > figures. After that we can discuss optimizations. There are no stats gathering of that sort possible with the code supplied in the kernel. You'd have to supply your own unfortunately. Gary -- Gary Palmer FreeBSD Core Team Member FreeBSD: Turning PC's into workstations. See http://www.FreeBSD.ORG/ for info To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message From owner-freebsd-fs Mon Oct 5 13:56:08 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id NAA00800 for freebsd-fs-outgoing; Mon, 5 Oct 1998 13:56:08 -0700 (PDT) (envelope-from owner-freebsd-fs@FreeBSD.ORG) Received: from panzer.plutotech.com (panzer.plutotech.com [206.168.67.125]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id NAA00770; Mon, 5 Oct 1998 13:55:53 -0700 (PDT) (envelope-from ken@panzer.plutotech.com) Received: (from ken@localhost) by panzer.plutotech.com (8.9.1/8.8.5) id OAA27658; Mon, 5 Oct 1998 14:55:38 -0600 (MDT) From: "Kenneth D. Merry" Message-Id: <199810052055.OAA27658@panzer.plutotech.com> Subject: Re: filesystem safety and SCSI disk write caching In-Reply-To: <199810051850.LAA23549@usr01.primenet.com> from Terry Lambert at "Oct 5, 98 06:50:04 pm" To: tlambert@primenet.com (Terry Lambert) Date: Mon, 5 Oct 1998 14:55:38 -0600 (MDT) Cc: oppermann@pipeline.ch, gpalmer@FreeBSD.ORG, freebsd-fs@FreeBSD.ORG X-Mailer: ELM [version 2.4ME+ PL28s (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org Terry Lambert wrote... > > > > > I can post (once again) the results of a Novell study on server usage > > > > > patterns. The 30,000 foot view for a typical server breaks down to: > > > > > > > > > > 75% reads > > > > > 15% writes > > > > > 8% directory search operations > > > > > 2% other > > > > > > I think that is very dependant on the server type. PC NetWare fileservers > > > probably have very different access patterns to (say) a web server or a mail > > > server. Let alone a news server. > > > > Is there a way to gather such statistics on FreeBSD? > > > > I'd like to run it on all my boxes (and others) to get representative > > figures. After that we can discuss optimizations. > > This would be a very good idea. > > It would be best to instrument at the lowest level, *below* where > soft updates does its implicit write gathering, and/or any elevator > sorting occurs. > > > I have suggested one set of instrumentation, which is interval sampling > of number of tagged commands outstanding, which would resolve once > and for all the current debate. Unfortunately, I don't have the > good firmware that Justin seems to have, and thus I can't really > make the measurements myself. It also seemed to me that the authors > of the CAM code would be in a better position about where to validly > determine max tag loading than I, and so would be better suited to > the task of coding up the statistics gathering. You can easily gather the statistics from the devstat code. The busy count will tell you how many transactions are oustanding for a particular device. The devstat code also keeps track of how many reads, writes, and other commands are sent to a drive. There's a program to dump out the statistics here: ftp://ftp.kdm.org/pub/FreeBSD/cam/ds.c I just updated it to grok the long generation size. (it was an int before) It doesn't matter for i386, but it does for the alpha. Ken -- Kenneth Merry ken@plutotech.com To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message From owner-freebsd-fs Mon Oct 5 16:07:07 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id QAA26916 for freebsd-fs-outgoing; Mon, 5 Oct 1998 16:07:07 -0700 (PDT) (envelope-from owner-freebsd-fs@FreeBSD.ORG) Received: from smtp04.primenet.com (smtp04.primenet.com [206.165.6.134]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id QAA26890 for ; Mon, 5 Oct 1998 16:06:55 -0700 (PDT) (envelope-from tlambert@usr04.primenet.com) Received: (from daemon@localhost) by smtp04.primenet.com (8.8.8/8.8.8) id QAA19060; Mon, 5 Oct 1998 16:06:42 -0700 (MST) Received: from usr04.primenet.com(206.165.6.204) via SMTP by smtp04.primenet.com, id smtpd019048; Mon Oct 5 16:06:39 1998 Received: (from tlambert@localhost) by usr04.primenet.com (8.8.5/8.8.5) id QAA28852; Mon, 5 Oct 1998 16:06:35 -0700 (MST) From: Terry Lambert Message-Id: <199810052306.QAA28852@usr04.primenet.com> Subject: Re: filesystem safety and SCSI disk write caching To: freebsd-fs@FreeBSD.ORG Date: Mon, 5 Oct 1998 23:06:35 +0000 (GMT) Cc: tlambert@primenet.com, oppermann@pipeline.ch In-Reply-To: <5889.907614151@gjp.erols.com> from "Gary Palmer" at Oct 5, 98 03:02:31 pm X-Mailer: ELM [version 2.4 PL25] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org > > > > > > 75% reads > > > > > > 15% writes > > > > > > 8% directory search operations > > > > > > 2% other > > > This would be a very good idea. > > > > It would be best to instrument at the lowest level, *below* where > > soft updates does its implicit write gathering, and/or any elevator > > sorting occurs. > > Depends on what you want to gather. The stats quotes above would seem to be > (IMHO) application<->FS layer transactions, not FS<->disk transactions. The FS is externalized via NCP, directly. The Novell stats are FS abstraction (NCP) to disk. You could argue that a network adapter was not a baseband connected disk controller, but Novell would disagree. The thing I would want instrumented, personally, is controller-to-disk transactions, with the intent of minimizing the number of transactions which occur. If the intent is to say that the disk is the bottleneck, then finding out how to minimize references to the bottleneck seems to be the way to go. This assumes that the disk is the bottleneck (which is why I suggested instrumenting the tagged command queueing to gather fill statistics on the tag queue. > And to be honest, you'd need both sets of figures (as well as a good > bit of analysis code which goes far beyond percentages) to obtain any > sort of `optimizations' based off those figures. Right. You'd need non-statistical profiling, probably using the instruction counter, and instrument on function entry and exit, as well as cycle elimination for elapsed time. While at Artisoft, we actually build a profiling system based on a Microsoft compiler flag for function entry instrumentation, coupled with stack hacks for function exit instrumentation (the compiler could only instrument entry) to gather this information. It was incidently on the Berkeley FFS in the Heidemann framework ported to Windows 95, ...including Soft Updates, implemented by Matt Day. We learned quite a few interesting things about the Windows 95 VM system, and VFAT page management. 8-(. > It would be really interesting to instrument at the application<->FS > layer a few of the different server types (POP3/MTA only, IMAP/MTA > only, news, etc) and see just what the access patterns are. I'm > personally very curious about the POP3/MTA systems using off-the-shelf > software. Actually, this is pretty trivial, using procfs and a wrapper script to initiate profiling. Sean has laid quite a bit of groundwork in this area (he recently got "truss" running on the Alpha, in fact). Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message From owner-freebsd-fs Fri Oct 9 21:25:18 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id VAA07571 for freebsd-fs-outgoing; Fri, 9 Oct 1998 21:25:18 -0700 (PDT) (envelope-from owner-freebsd-fs@FreeBSD.ORG) Received: from federation.addy.com (federation.addy.com [207.239.68.2]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id VAA07387 for ; Fri, 9 Oct 1998 21:24:29 -0700 (PDT) (envelope-from francisco@natserv.com) Received: from quisqueya.natserv.com (TC1-dial-58-142.oldslip.inch.com [207.240.142.58]) by federation.addy.com (8.8.5/8.6.12) with ESMTP id AAA21260 for ; Sat, 10 Oct 1998 00:24:03 -0400 (EDT) Message-Id: <199810100424.AAA21260@federation.addy.com> X-Mailer: XFMail 1.3 [p0] on FreeBSD X-Priority: 3 (Normal) Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 8bit MIME-Version: 1.0 Date: Sat, 10 Oct 1998 00:24:30 -0000 (GMT) Reply-To: francisco@natserv.com From: Francisco Reyes To: freebsd-fs@FreeBSD.ORG Subject: Optimizing space utilization Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org I am looking for certain info on the filesystem to try to maximize space utilization. Any pointers/URLs would be appreciated. What I am trying to find out is: -- What is the smallest unit FreeBSD manages. Is this a block or fragment? >From what I have seen FreeBSD creates blocks of size 8K by default. -- What is a fragment (man newfs has a "frag size") --What space does an inode entry takes? How many file entries go into each Inode? If I know how many files I have for a fs I am trying to see if space could be saved by using less inodes. -- One of the options in newfs is to optimize for "space". Does it really save space? What kind of degradation will be seen by using that flag instead of "time" optimization?. I will use the info for a small server I am setting, but will also put it into a small tutorial I am writing: "Sizing filesystems" ---- francisco@natserv.com The power to serve. http://www.freebsd.org To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message From owner-freebsd-fs Fri Oct 9 23:07:55 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id XAA22027 for freebsd-fs-outgoing; Fri, 9 Oct 1998 23:07:55 -0700 (PDT) (envelope-from owner-freebsd-fs@FreeBSD.ORG) Received: from allegro.lemis.com (allegro.lemis.com [192.109.197.134]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id XAA22022 for ; Fri, 9 Oct 1998 23:07:51 -0700 (PDT) (envelope-from grog@freebie.lemis.com) Received: from freebie.lemis.com (freebie.lemis.com [192.109.197.137]) by allegro.lemis.com (8.9.1/8.9.0) with ESMTP id PAA10654; Sat, 10 Oct 1998 15:37:32 +0930 (CST) Received: (from grog@localhost) by freebie.lemis.com (8.9.1/8.9.0) id PAA20445; Sat, 10 Oct 1998 15:37:26 +0930 (CST) Message-ID: <19981010153725.T3369@freebie.lemis.com> Date: Sat, 10 Oct 1998 15:37:25 +0930 From: Greg Lehey To: francisco@natserv.com, freebsd-fs@FreeBSD.ORG Subject: Re: Optimizing space utilization References: <199810100424.AAA21260@federation.addy.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.91.1i In-Reply-To: <199810100424.AAA21260@federation.addy.com>; from Francisco Reyes on Sat, Oct 10, 1998 at 12:24:30AM -0000 WWW-Home-Page: http://www.lemis.com/~grog Organization: LEMIS, PO Box 460, Echunga SA 5153, Australia Phone: +61-8-8388-8286 Fax: +61-8-8388-8725 Mobile: +61-41-739-7062 Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org On Saturday, 10 October 1998 at 0:24:30 -0000, Francisco Reyes wrote: > I am looking for certain info on the filesystem to try to maximize space > utilization. > > Any pointers/URLs would be appreciated. > > What I am trying to find out is: > > -- What is the smallest unit FreeBSD manages. Is this a block or fragment? A fragment. > From what I have seen FreeBSD creates blocks of size 8K by default. > > -- What is a fragment (man newfs has a "frag size") The smallest unit FreeBSD manages. > --What space does an inode entry takes? Currently, 260 bytes. > How many file entries go into each Inode? One. The inode *is* the file; the names are just links to the inode. > If I know how many files I have for a fs I am trying to see if space > could be saved by using less inodes. None. You need exactly one inode per file. > -- One of the options in newfs is to optimize for "space". Does it really > save space? What kind of degradation will be seen by using that flag > instead of "time" optimization?. The simple (and useless :-) answer is "Time optimization saves time; space optimization saves space". It's difficult to summarize the differences in a sentence or two. "The Design and Implementation of the 4.3BSD UNIX Operating System" goes into some detail about this. There's also McKusick's paper on UFS, which you'll find on the second CD-ROM in /usr/share/doc/smm/05.fastfs/. Greg -- See complete headers for address, home page and phone numbers finger grog@lemis.com for PGP public key To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message From owner-freebsd-fs Sat Oct 10 19:50:49 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id TAA01804 for freebsd-fs-outgoing; Sat, 10 Oct 1998 19:50:49 -0700 (PDT) (envelope-from owner-freebsd-fs@FreeBSD.ORG) Received: from federation.addy.com (federation.addy.com [207.239.68.2]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id TAA01793 for ; Sat, 10 Oct 1998 19:50:41 -0700 (PDT) (envelope-from francisco@natserv.com) Received: from quisqueya.natserv.com (TC4-dial-177-142.oldslip.inch.com [207.240.142.177]) by federation.addy.com (8.8.5/8.6.12) with ESMTP id WAA20883; Sat, 10 Oct 1998 22:50:19 -0400 (EDT) Message-Id: <199810110250.WAA20883@federation.addy.com> X-Mailer: XFMail 1.3 [p0] on FreeBSD X-Priority: 3 (Normal) Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 8bit MIME-Version: 1.0 In-Reply-To: <19981010153725.T3369@freebie.lemis.com> Date: Sat, 10 Oct 1998 22:50:47 -0000 (GMT) Reply-To: francisco@natserv.com From: Francisco Reyes To: Greg Lehey Subject: Re: Optimizing space utilization Cc: freebsd-fs@FreeBSD.ORG Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org On 10-Oct-98 Greg Lehey wrote: >> If I know how many files I have for a fs I am trying to see if space >> could be saved by using less inodes. > > None. You need exactly one inode per file. I didn't phrase that question properly. What I meant was: if I do a newfs and create less inodes, will I save any space. Today I was able to check. On a 1 Gig partition by using 1/4 the number of inodes (sent -i 16384 to newfs) I was able to save a little over 30MB. A 3% saving. :-) I was still left with over 64K inodes, which I doubt I will use. Thanks for the reference to the paper in /usr/share/doc/...fastfs Also found another paper in the "papers" directory which relates to performance. Those should make good subway reading. :-) ---- francisco@natserv.com To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message From owner-freebsd-fs Sat Oct 10 19:55:15 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id TAA02823 for freebsd-fs-outgoing; Sat, 10 Oct 1998 19:55:15 -0700 (PDT) (envelope-from owner-freebsd-fs@FreeBSD.ORG) Received: from allegro.lemis.com (allegro.lemis.com [192.109.197.134]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id TAA02788 for ; Sat, 10 Oct 1998 19:55:06 -0700 (PDT) (envelope-from grog@freebie.lemis.com) Received: from freebie.lemis.com (freebie.lemis.com [192.109.197.137]) by allegro.lemis.com (8.9.1/8.9.0) with ESMTP id MAA13659; Sun, 11 Oct 1998 12:24:48 +0930 (CST) Received: (from grog@localhost) by freebie.lemis.com (8.9.1/8.9.0) id MAA23208; Sun, 11 Oct 1998 12:24:47 +0930 (CST) Message-ID: <19981011122446.P3369@freebie.lemis.com> Date: Sun, 11 Oct 1998 12:24:46 +0930 From: Greg Lehey To: francisco@natserv.com Cc: freebsd-fs@FreeBSD.ORG Subject: Re: Optimizing space utilization References: <19981010153725.T3369@freebie.lemis.com> <199810110250.WAA20883@federation.addy.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.91.1i In-Reply-To: <199810110250.WAA20883@federation.addy.com>; from Francisco Reyes on Sat, Oct 10, 1998 at 10:50:47PM -0000 WWW-Home-Page: http://www.lemis.com/~grog Organization: LEMIS, PO Box 460, Echunga SA 5153, Australia Phone: +61-8-8388-8286 Fax: +61-8-8388-8725 Mobile: +61-41-739-7062 Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org On Saturday, 10 October 1998 at 22:50:47 -0000, Francisco Reyes wrote: > > On 10-Oct-98 Greg Lehey wrote: >>> If I know how many files I have for a fs I am trying to see if space >>> could be saved by using less inodes. >> >> None. You need exactly one inode per file. > > I didn't phrase that question properly. > What I meant was: if I do a newfs and create less inodes, will I save any > space. Yes. > Today I was able to check. > On a 1 Gig partition by using 1/4 the number of inodes (sent -i 16384 to > newfs) I was able to save a little over 30MB. A 3% saving. :-) > I was still left with over 64K inodes, which I doubt I will use. Interesting. At 262 bytes per inode, I'd only expect a saving of about 12 MB. > Thanks for the reference to the paper in /usr/share/doc/...fastfs > Also found another paper in the "papers" directory which relates to > performance. You're welcome > Those should make good subway reading. :-) It would be interesting what people looking over your shoulder would think :-) Greg -- See complete headers for address, home page and phone numbers finger grog@lemis.com for PGP public key To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message From owner-freebsd-fs Sat Oct 10 20:04:30 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id UAA03755 for freebsd-fs-outgoing; Sat, 10 Oct 1998 20:04:30 -0700 (PDT) (envelope-from owner-freebsd-fs@FreeBSD.ORG) Received: from federation.addy.com (federation.addy.com [207.239.68.2]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id UAA03750 for ; Sat, 10 Oct 1998 20:04:28 -0700 (PDT) (envelope-from francisco@natserv.com) Received: from quisqueya.natserv.com (TC4-dial-177-142.oldslip.inch.com [207.240.142.177]) by federation.addy.com (8.8.5/8.6.12) with ESMTP id XAA22392; Sat, 10 Oct 1998 23:04:08 -0400 (EDT) Message-Id: <199810110304.XAA22392@federation.addy.com> X-Mailer: XFMail 1.3 [p0] on FreeBSD X-Priority: 3 (Normal) Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 8bit MIME-Version: 1.0 In-Reply-To: <19981011122446.P3369@freebie.lemis.com> Date: Sat, 10 Oct 1998 23:04:35 -0000 (GMT) Reply-To: francisco@natserv.com From: Francisco Reyes To: Greg Lehey Subject: Re: Optimizing space utilization Cc: freebsd-fs@FreeBSD.ORG Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org On 11-Oct-98 Greg Lehey wrote: >> On a 1 Gig partition by using 1/4 the number of inodes (sent -i 16384 >> to newfs) I was able to save a little over 30MB. A 3% saving. :-) >> I was still left with over 64K inodes, which I doubt I will use. > Interesting. At 262 bytes per inode, I'd only expect a saving of > about 12 MB. Something else I noticed.. The total space reported by df (heading 1K-blocks) also went up slightly. Don't recall exact number, but I think it was 3 to 5MB. ---- francisco@natserv.com To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message