Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 8 Dec 2008 10:50:32 -0600
From:      Doug Poland <doug@polands.org>
To:        Ivan Voras <ivoras@freebsd.org>
Cc:        freebsd-questions@freebsd.org
Subject:   Re: Backup complete gmirror/gstripe/gjournal drives, how-to?
Message-ID:  <20081208165032.GA99675@polands.org>
In-Reply-To: <9bbcef730812080822q495da820h1fdd28c9ab2c96d7@mail.gmail.com>
References:  <493AFD81.6020705@polands.org> <ghipe0$dm5$1@ger.gmane.org> <20081208153348.GA57784@polands.org> <9bbcef730812080822q495da820h1fdd28c9ab2c96d7@mail.gmail.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, Dec 08, 2008 at 05:22:05PM +0100, Ivan Voras wrote:
> 2008/12/8 Doug Poland <doug@polands.org>:
> > On Mon, Dec 08, 2008 at 10:29:36AM +0100, Ivan Voras wrote:
> 
> >> I hope you understand the problems with this kind of backup
> >> procedures.
> >>
> > Actually, I don't, that's why I'm posting :)  My intent is to have a
> 
> Here are a couple of things:
> 
I'd like to reiterate a statement I made in my OP, this is a "one-time"
backup to preserve GEOM metadata.  After this, I will use dump and rsync
as I ordinarily do.

> a) How do you intend to restore the images? 
>
Booting from a live CD and reversing the dd netcat commands.  Something
like:

livecd# nc -l 1234 | dd of=/dev/ad4
backup# gzip -dc /data/backup/ad4.img | nc livecd 1234

> You need drives with >= size of the original drives (and images). If
> you have larger drives, the extra space might be wasted.
>
Understood

> b) With this method you cannot use (or actually, you can but it's very
> hard) any kind of differential/incremental backup, and in addition to
> the used space you're also backuping the unused/"empty" space.
>
My intent would be to use the dd images to restore the disk partitions
and GEOM metadata, then I would use restore to get the latest data.

> c) dd as you've used it will skip bad sectors, in a bad way - it will
> either stop (according to the documentation) or concatenate data
> blocks around the ones containing the bad sector (e.g. if a sector in
> the tenth block of 2 MB is bad, your file will contain blocks
> ...8,9,11,12,...). See conv=noerror,sync options to dd.
>
Thank you for bringing that out.  I have seen references to those dd
options while googling this topic.  Will using conv=noerror,sync
preserve the disk in a viable state?

Fundamentally, is there a better way to accomplish what I'm trying to
do?  Basically a bare-metal backup/restore of a 4-disk RAID-10 built
using gmirror/gstripe with gjournal on top.  If my dd solution is too
problematic, I'll do something else.

Your comments and suggestions are very welcome...

-- 
Regards,
Doug



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20081208165032.GA99675>