From owner-freebsd-hackers Sun Mar 31 22:35:42 1996 Return-Path: owner-hackers Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id WAA03483 for hackers-outgoing; Sun, 31 Mar 1996 22:35:42 -0800 (PST) Received: from news1.gtn.com (news1.gtn.com [192.109.159.3]) by freefall.freebsd.org (8.7.3/8.7.3) with ESMTP id WAA03478 for ; Sun, 31 Mar 1996 22:35:40 -0800 (PST) Received: (from uucp@localhost) by news1.gtn.com (8.7.2/8.7.2) id IAA13077; Mon, 1 Apr 1996 08:15:19 +0200 (MET DST) Received: from localhost (localhost [127.0.0.1]) by gun.de (8.7.5/8.7.3) with SMTP id IAA00544; Mon, 1 Apr 1996 08:04:49 +0200 (MET DST) Date: Mon, 1 Apr 1996 08:04:47 +0200 (MET DST) From: Andreas Klemm To: Joerg Wunsch cc: FreeBSD hackers Subject: Re: Lowering minfree to 1% on large disks In-Reply-To: <199603312034.WAA00828@uriah.heep.sax.de> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-hackers@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk -----BEGIN PGP SIGNED MESSAGE----- On Sun, 31 Mar 1996, J Wunsch wrote: > As Brian Tao wrote: > > > I know the tunefs man page contains warnings about lowering the > > minfree threshold on a disk to below 5%, but besides file write > > performance, is there any other reason *not* to drop it down to 1 or > > 2 percent? > > File *write* performance? I think it's the overall file system > performance. Read the daemon book... When this 10% rule came out, you had to deal with disk sizes about 100 MB and such ... 10% were about 10 MB. Now you have 4 GB disks and 10% are about 400MB. More space than those disks ever had. I think the 10% rule is too static. Shouldn't that be adapted in relation to the disk or better said filesystem size ?! 100 MB should be enough for a 4GB partition ... Instead of... 100 MB - 10% - 10 MB 200 MB - 10% - 20 MB 500 MB - " - 50 MB 1000 MB - " - 100 MB Perhaps 100 MB - xx% - 10 MB 200 MB - xx% - 15 MB 500 MB - " - 30 MB 1000 MB - " - 70 MB Very very estimated ;-) > Perhaps for a rather static file system, where you're keeping the > fragmentation low by restore(8)ing the file system frequently after > modifications, it might be okay, or for file systems that are only > rarely used at all. Or does this mean an impact on write performance, since each cylinder group needs a certain amount of space ???? If this would be the case, then 10% at all would again make sense... One should have disk and time to verify that with bonnie... Andreas /// - -- andreas@knobel.gun.de /\/\___ Wiechers & Partner Datentechnik GmbH Andreas Klemm ___/\/\/ $$ Support Unix - aklemm@wup.de $$ pgp p-key http://www-swiss.ai.mit.edu/~bal/pks-toplev.html >>> powered by <<< ftp://sunsite.unc.edu/pub/Linux/system/Printing/aps-491.tgz >>> FreeBSD <<< -----BEGIN PGP SIGNATURE----- Version: 2.6.2 iQCVAwUBMV9yAPMLpmkD/U+FAQGCYQP8DNeUouWc20J3UzA1TBnIhaIsfPpwkMa3 wRMQrS2coxpq6EysrzAlwoC/6s+uBNTNaYCgWjozVSNhiw9nYOt8cAw2ViDHRkpZ XERo4ae4yh2P1iWLFmK5QVz1tXQ3CWhIBctPVj1mILWVVeb1Ymn+mhBmsPe5P8lW RduzDd9naqc= =5YHI -----END PGP SIGNATURE-----