Date: Thu, 15 Feb 2007 00:38:48 +0100 From: Pawel Jakub Dawidek <pjd@FreeBSD.org> To: Eric Anderson <anderson@centtech.com> Cc: "freebsd-fs@freebsd.org" <freebsd-fs@freebsd.org> Subject: Re: fsck times/memory sizes/etc Message-ID: <20070214233848.GA13680@garage.freebsd.pl> In-Reply-To: <45D321F8.2000808@centtech.com> References: <45D321F8.2000808@centtech.com>
next in thread | previous in thread | raw e-mail | index | archive | help
--+QahgC5+KEYLbs62 Content-Type: text/plain; charset=iso-8859-2 Content-Disposition: inline Content-Transfer-Encoding: quoted-printable On Wed, Feb 14, 2007 at 08:51:36AM -0600, Eric Anderson wrote: > I just did some quick playing around, doing newfs and fsck's on fresh fil= e systems. I did one set on a 1Gb malloc'ed md, and another on a 65GB real= disk device. The disk=20 > was only being used for this test, nothing else. >=20 > I don't claim these numbers are perfect, but I did run each test 3-4 time= s to make sure they were consistent. >=20 > I found it interesting that the fsck times didn't reduce once all the fil= es/directories were deleted. This may be because of UFS2 has lazy inodes allocator - it doesn't initialize all inode blocks at newfs time, so fsck verifies only allocated inode blocks, thus runs faster on new file system. When you fill your file system, inode blocks are allocated, but are not reclaimed when you delete file/directories. You may try the same tests with UFS1, which allocates all inode blocks at newfs time. --=20 Pawel Jakub Dawidek http://www.wheel.pl pjd@FreeBSD.org http://www.FreeBSD.org FreeBSD committer Am I Evil? Yes, I Am! --+QahgC5+KEYLbs62 Content-Type: application/pgp-signature Content-Disposition: inline -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (FreeBSD) iD8DBQFF052IForvXbEpPzQRAnu8AKDPcMJel6A9qtndNIcDJCRWQdJLRwCePZt4 g8uPLFERBVPnTEvkfQU7/2w= =2sib -----END PGP SIGNATURE----- --+QahgC5+KEYLbs62--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20070214233848.GA13680>