From owner-freebsd-geom@FreeBSD.ORG Tue Jan 2 19:40:54 2007 Return-Path: X-Original-To: freebsd-geom@freebsd.org Delivered-To: freebsd-geom@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id 29CEA16A415 for ; Tue, 2 Jan 2007 19:40:54 +0000 (UTC) (envelope-from rick@kiwi-computer.com) Received: from kiwi-computer.com (keira.kiwi-computer.com [63.224.10.3]) by mx1.freebsd.org (Postfix) with SMTP id A824F13C441 for ; Tue, 2 Jan 2007 19:40:53 +0000 (UTC) (envelope-from rick@kiwi-computer.com) Received: (qmail 84211 invoked by uid 2001); 2 Jan 2007 19:46:31 -0000 Date: Tue, 2 Jan 2007 13:46:31 -0600 From: "Rick C. Petty" To: Ivan Voras Message-ID: <20070102194631.GB83431@keira.kiwi-computer.com> References: <975053160612310913t3dadcc02yfac58f6fbf0a49df@mail.gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.4.2.1i Cc: freebsd-geom@freebsd.org Subject: Re: Recommended gmirror solution with swap? X-BeenThere: freebsd-geom@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list Reply-To: rick-freebsd@kiwi-computer.com List-Id: GEOM-specific discussions and implementations List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 02 Jan 2007 19:40:54 -0000 On Mon, Jan 01, 2007 at 07:07:59PM +0100, Ivan Voras wrote: > > There's no technical problem with it, it's more a matter of organization > and convenience. For example: you don't really need the swap to survive > a crash, so there's no need to introduce the overhead of mirroring it - > better to leave 2x the space configured as swap area. If you have more > than one area, the system should take advantage of it and use both to > maximize speed (in theory this should work as an implicit RAID0 across > the swap areas). Mirror the whole disk, but instead of having /etc/fstab use /dev/mirror/gm0s1b as a swap device, have it use both /dev/mirror/ad0s1b and /dev/mirror/ad1s1b.. Since gm0s1b isn't used, there shouldn't be a problem referencing the underlying device nodes =) I'm sure the current implementation does not allow for this, but I can't see any reason why this shouldn't be *allowed*. However, it may not be *possible* because AFAIK there's no way in GEOM for bsdlabel to tell its provider (gmirror) that certain portions aren't in use, so the whole devices (ad0 and ad1) are being fully consumed. An interesting idea though... -- Rick C. Petty