Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 11 Oct 2004 00:25:46 +0200
From:      Christian Hiris <4711@chello.at>
To:        freebsd-current@freebsd.org
Cc:        Chris Elsworth <chris@shagged.org>
Subject:   Re: RAID1 with gmirror
Message-ID:  <200410110026.03777.4711@chello.at>
In-Reply-To: <20041010165852.GA15854@shagged.org>
References:  <20041010110159.GA91160@pcs28.suedfactoring.de> <200410101522.45548.4711@chello.at> <20041010165852.GA15854@shagged.org>

next in thread | previous in thread | raw e-mail | index | archive | help
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Sunday 10 October 2004 18:58, Chris Elsworth wrote:
> On Sun, Oct 10, 2004 at 03:22:28PM +0200, Christian Hiris wrote:

> > # gmirror configure -v -b load mirror0 ad0
>             ^^^^^^^^^
>            label?

Yes, that should read "label" - sorry! 

> I'm quite interested that you didn't bother re-labelling the new
> mirror device? I always thought that you had to (and I have been doing
> this):

No, re-labeling wasn't necessary. gmirror could read the existing labels and 
created the corresponding mirror*s** devices showed up under /dev/mirror/. 
fsck also went fine. I think this should work, as gmirror stores it's 
metadata in the last sector of it's providers.

> 4) bsdlabel -e ad0 [taking into account the new 16 byte offset]

I never read about the new offset - until now I had no time to study the 
gmirror code. Where did you find this interresting information?

> Is this all completely unnecessary? The first few times I played with
> gmirror, I was doing something similar to your method, but if you do
> bsdlabel -r /dev/mirror/mirror0s1 (or whatever) on your setup, don't
> you get warnings about partition sizes?

Yes, I get those warnings, too - but I decided to leave the labels as is, 
because on a test array I found this working w/o any datacorruption.

An example in 'man gmirror' describes how to set up a gmirror array on a disk 
with valid data. It uses 'gmirror lable' w/o extra labeling. I understand it 
the way that the term 'valid data' means 'existing (non-gmirror-)data on an 
already (before-gmirror-)labeled disk'. Maybe my understanding is wrong and 
you gave me the hint to the reason why my mirrors break on startup. I will do 
some testing on this.

The 2nd reason why I didn't touch my old labels, is that I want to be able to 
step back to ataraid as simple as possible, just in case of a breakage of the  
gmirror code.

> That's what worried me enough 
> to go through all the rigmarole above, which fixes those warnings.

I use gmirror since the day before yesterday, so for now there are still some 
things that I couldn't test and possibly I overlooked some things. At a first 
look, my experiences were not so bad, I lost no data and gained a lot of 
flexibility in my hdd subsystem.

- -- 
Christian Hiris <4711@chello.at> | OpenPGP KeyID 0x3BCA53BE 
OpenPGP-Key at hkp://wwwkeys.eu.pgp.net and http://pgp.mit.edu
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.6 (FreeBSD)

iD8DBQFBabb709WjGjvKU74RAjzIAJ4p+4RUbBgYLGeJlQX/GxXrvR1dswCdFOSy
K5fuNpfvidRxvSEYVRYAYOY=
=gXBY
-----END PGP SIGNATURE-----



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200410110026.03777.4711>