Date: Fri, 2 Mar 2007 11:12:25 -0500 From: Jerry McAllister <jerrymc@msu.edu> To: Ivan Voras <ivoras@fer.hr> Cc: freebsd-questions@freebsd.org Subject: Re: defrag Message-ID: <20070302161225.GB90036@gizmo.acns.msu.edu> In-Reply-To: <es7tvd$b33$1@sea.gmane.org> References: <539c60b90703010849x33dd4bbbt8f6ca6aa0c8e83a0@mail.gmail.com> <20070301192109.A24369@chylonia.3miasto.net> <20070302085100.125cf488@localhost> <20070301221738.GA86154@gizmo.acns.msu.edu> <es7tvd$b33$1@sea.gmane.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, Mar 02, 2007 at 02:17:31AM +0100, Ivan Voras wrote: > Jerry McAllister wrote: > > > Well, it would do some, but for the greatest effect, you would need: > > dump + rm -rf * + restore > > This is nitpicking so ignore it: deleting all files on UFS2 volume won't > restore it to it's pristine state because inodes are lazily initialized. > It doesn't have anything to do with fragmentation, but will make fsck > run a little longer. > True it wouldn't be quite pristine because files would have different inodes assigned when they get reloaded than they might have if it was newfs-ed before reloading. That might make fsck run a tiny bit slower. But it wouldn't be any difference for a running system file access. On the other hand, doing all this either way wouldn't make any difference in performance for file access in a running system because so-called fragmentation is not an issue in the UNIX file system - except in the small possibility that it might make a bit of difference in a file system filled to capacity, well in to the reserve where non-root processes are not allowed to write anyway. I don't know just how close to absolutely full you have to get to see any difference, but it is beyond what users would normally get to. ////jerry
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20070302161225.GB90036>