Date: Tue, 01 May 2007 08:03:24 -0500 From: Eric Anderson <anderson@freebsd.org> To: Frans Haarman <F.Haarman@giessen.nl> Cc: freebsd-geom@freebsd.org Subject: Re: gmirror magic ? Message-ID: <46373A9C.8020106@freebsd.org> In-Reply-To: <2DC959620A73E842969792F5B47FCA01037D44DC@dg-exch1.giessen.nl> References: <2DC959620A73E842969792F5B47FCA01037D44DC@dg-exch1.giessen.nl>
next in thread | previous in thread | raw e-mail | index | archive | help
On 05/01/07 07:24, Frans Haarman wrote: > Can someone explain this to me ? I was testing gmirror/gstripe and > decided to > remove the vnode files stripe0 & stripe1. This resulted in a mirror > which kept on > working..... > > > /dev/mirror/mirro0 on /mnt (ufs, local) > > DEVEL# ls /mnt/ > .snap anders test testje > > DEVEL# ls /data/STRIPING/ > create-image-file mnt mnt0 > mnt1 > > > DEVEL# mdconfig -l -u md1 > md1 vnode 512M /data/STRIPING/stripe0 > DEVEL# mdconfig -l -u md2 > md2 vnode 512M /data/STRIPING/stripe1 > DEVEL# mdconfig -l -u md123 > md123 vnode 512M /data/STRIPING/stripe1 Does that look right to you? You have 2 md devices using the same file? > DEVEL# gmirror status > Name Status Components > mirror/mirro0 COMPLETE md2 > md123 Where exactly does the gstripe come in? > DEVEL# dd if=/dev/zero of=where-does-this-go bs=1m count=512 I'm assuming your cwd was /mnt? > /mnt: write failed, filesystem is full > dd: where-does-this-go: No space left on device > 431+0 records in > 430+0 records out > 450887680 bytes transferred in 67.178881 secs (6711747 bytes/sec) > DEVEL# Ok - looks about right to me. You had a 512M mirror, and you wrote enough to fill the file system up.. ? Right? > This seems quite strange to me :) Its probably the way vnode disks are > attached > to the filesystem ? I'm missing something. You might try to paint a bit clearer picture, as it's pretty hard to understand what exactly we should be looking for.. Eric
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?46373A9C.8020106>