Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 12 Jul 2003 17:59:04 +1000
From:      Peter Jeremy <PeterJeremy@optushome.com.au>
To:        Joshua Oreman <oremanj@www.get-linux.org>
Cc:        hackers@freebsd.org
Subject:   Re: Mirroring using vinum+NFS
Message-ID:  <20030712075903.GA4105@cirb503493.alcatel.com.au>
In-Reply-To: <20030710203821.GA5576@webserver.get-linux.org>
References:  <20030710201013.GA1892@cirb503493.alcatel.com.au> <20030710203821.GA5576@webserver.get-linux.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, Jul 10, 2003 at 01:38:21PM -0700, Joshua Oreman wrote:
>On Fri, Jul 11, 2003 at 06:10:13AM +1000 or thereabouts, Peter Jeremy wrote:
>> Consider system A as the server and system B as the mirror.  In theory,
>> on system A I should be able to:
>> 	mount B:/big/data/blob /remote
>> 	vnconfig /big/data/block as vn0
>> 	vnconfig /remote as vn1
>> 	vinum mirror vn0+vn1 as serverdata
>> 	mount /dev/vinum/serverdata /serverdata
>> 	export /serverdata
>
>This will not help you at all.
>Vnconfig only deals with *files*. Directories are outside its realm.

I suggest you re-read vnconfig(8).  vnconfig turns a file (local or
NFS mounted) into a device which can then be turned into a filesystem
- either directly or via a LVM (eg vinum).

>Since you have no direct write permission to the device over NFS, Vinum
>has no chance of working there.

There's no reference to remotely accessing a device.  The above
actually works perfectly in the case where both systems are up (modulo
using correct pathnames, some disklabeling and a newfs).  I think I'll
need to use 'mount -s' for the first mount to make it robust against
system B failing.

Peter



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20030712075903.GA4105>