From owner-freebsd-chat Wed Jul 14 19:37:12 1999 Delivered-To: freebsd-chat@freebsd.org Received: from smtp02.primenet.com (smtp02.primenet.com [206.165.6.132]) by hub.freebsd.org (Postfix) with ESMTP id 656C11547F for ; Wed, 14 Jul 1999 19:37:11 -0700 (PDT) (envelope-from tlambert@usr07.primenet.com) Received: (from daemon@localhost) by smtp02.primenet.com (8.8.8/8.8.8) id TAA13238; Wed, 14 Jul 1999 19:35:28 -0700 (MST) Received: from usr07.primenet.com(206.165.6.207) via SMTP by smtp02.primenet.com, id smtpd013210; Wed Jul 14 19:35:25 1999 Received: (from tlambert@localhost) by usr07.primenet.com (8.8.5/8.8.5) id TAA12574; Wed, 14 Jul 1999 19:35:23 -0700 (MST) From: Terry Lambert Message-Id: <199907150235.TAA12574@usr07.primenet.com> Subject: Re: Known MMAP() race conditions ... ? To: regnauld@ftf.net (Phil Regnauld) Date: Thu, 15 Jul 1999 02:35:23 +0000 (GMT) Cc: davids@webmaster.com, chat@FreeBSD.ORG In-Reply-To: <19990714204622.17443@ns.int.ftf.net> from "Phil Regnauld" at Jul 14, 99 08:46:22 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-chat@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org > FFS is even better: it doesn't fragment. Sort of. It fragments after reaching better than 85% fill, and the more over this it goes, the more it fragments. In addition, although it's relatively trivial to expand a partition size (either by editing the disklabel to use contiguous free space, or by using Vinum or ccd to define a larger logical drive), doing so will not cause the simple hash to prefer the new area until it reaches an equivalent fill. The result is that, if the total fill in the original (now sub-) region exceeds 85%, that region suffers increased fragmentation. This means that if you have a 3G FS and you expand it to a 4G FS, you should expect significant fragmentation. The generally recommended "fix" for this is backup, then restore, the files, but a "defragmenter" that caused the data to be rehashed forcefully so that all cylinder groups has equivalent fill would be a better approach. Similarly, you can't reduce the size of an FFS partition. A defragmenter that could be told "this zone is off limits" would also be useful for shrinking FFS partitions. 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-chat" in the body of the message