Date: Thu, 14 Dec 2006 23:56:54 +0100 From: Ulf Lilleengen <lulf@stud.ntnu.no> To: Olivier Cochard-Labbe <olivier@freenas.org> Cc: freebsd-geom@freebsd.org Subject: Re: geom raid5 module Message-ID: <20061214225654.GA14991@vimes.studby.ntnu.no> In-Reply-To: <3131aa530612130145l77ebb6f6ga7d4ac54716b5cc0@mail.gmail.com> References: <3131aa530612110522y7b8ccad5k9d5735277b4d5ffa@mail.gmail.com> <20061211164844.GA3827@vimes.studby.ntnu.no> <3131aa530612130145l77ebb6f6ga7d4ac54716b5cc0@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On ons, des 13, 2006 at 10:45:14am +0100, Olivier Cochard-Labbe wrote: > Here is, for example, the first problem problem with geom vinum RAID5: > > I have 4 disks: > - ad0: 203MB > - ad1: 101MB > - ad2: 306MB > - ad3: 101MB > > If I create a raid 5 volume with these disk, can you confirme that the > theorical size should be 300MB (4 disks of the smallest drive: 100MB) > ? It's the size given with graid5 and with RAID caclulator > (http://dorm.tunkeymicket.com/raidcalc/). > But with geom vinum I obtain a 611MB disk: > Hi, I found the "bug"! The thing is that gvinum sets the plex-size after the last subdisk created instead of the smallest. I started to make a fix to make it set it to the lowest, but then I began thinking. You should not make different-sized subdisks in a raid5 configuration. Then I discovered this is "fixed" in CURRENT, by giving you an error when you try to create a raid5 volume with different-sized subdisks. And I think this makes sense. If you create a raid5-volume, all subdisks should be the same size. However, I'm thinking of maybe make gvinum be more "autodetectable" on this upon creation by actually examining which subdisk is smallest when you give it a raid5 plex with different-sized subdisks. I'll look more into your second problem tomorrow :) -- Ulf Lilleengen
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20061214225654.GA14991>