[PC-BSD Testing] Backing up a PC-BSD disk

Odhiambo Washington odhiambo at gmail.com
Mon Nov 1 08:15:22 PDT 2010


On Mon, Nov 1, 2010 at 6:00 PM, Kris Moore <kris at pcbsd.org> wrote:

> On Mon, Nov 01, 2010 at 05:39:33PM +0300, Odhiambo Washington wrote:
> > On Mon, Nov 1, 2010 at 5:28 PM, Kris Moore <kris at pcbsd.org> wrote:
> >
> > > On Mon, Nov 01, 2010 at 12:49:47PM +0300, Odhiambo Washington wrote:
> > > > I installed PC-BSD (FreeBSD option) on a disk on which I only created
> /
> > > > (ufs+journal) and swap.
> > > > >From the handbook (
> > > > http://www.freebsd.org/doc/en/articles/gjournal-desktop/article.html),
> I
> > > am
> > > > still unable to tell how much space the PC-BSD installer allocated
> for
> > > the
> > > > journal provider, which according to my understanding, resides on the
> > > same
> > > > partition as the data.
> > > >
> > > > Is the journal size defaulting to 1GB as the handbook says, or does
> the
> > > > installer use a different value?
> > > >
> > > >
> > > > [wash at gw ~]$ df -h
> > > > Filesystem          Size   Used     Avail    Capacity Mounted on
> > > > /dev/label/rootfs0  273G    9.8G   242G   4%         /
> > > > devfs                   1.0K     1.0K       0B   100%     /dev
> > > > devfs                   1.0K     1.0K       0B   100%
> /var/named/dev
> > > >
> > > > I am so much used UFS2 type filesystem where I create / and swap and
> I'd
> > > > always get an identical size disk for backup. The backup would take
> the
> > > > form:
> > > >
> > > > dump -L0af - / | (cd /disk2; restore -rf -) # And this is disk
> > > duplication
> > > > to me, so I can use disk2 as primary disk should disk one fail.
> > > >
> > > > When it comes to ufs+journal, I am a bit lost on how to do things
> right
> > > when
> > > > I need to make a second disk (disk2) to have same partition layout as
> > > disk
> > > > 1.
> > > >
> > > > Suppose both disks are 320GB, can I use dd to create the duplicate
> > > > structure, with journaling enabled??
> > > > dd if=/dev/adX of=/dev/adY bs=1m
> > > >
> > > > But if you tell me that the PC-BSD installer allocated 1GB for the
> > > journal,
> > > > then I can follow the handbook instructions to partition my disk
> > > > accordingly. However, given that the installer mounted
> /dev/label/rootfs0
> > > on
> > > > /, I am not sure what will be the name of my filesystem on disk2 when
> I
> > > > create it from the CLI. I'd love to partition as follows:
> > > >
> > > > gw# bsdlabel -A /dev/ad4
> > > > bsdlabel: /dev/ad4: no valid label found
> > > >
> > > > Since I am unable to read the disk label of the first disk, I am not
> > > certain
> > > > how to go about creating a similar label on disk2. I thought I could
> read
> > > > out the disk label on disk1 and just write that to disk 2.
> > > >
> > > >
> > > > Secondly, I am not sure how I can backup the first disk onto disk2
> using
> > > > dump once I have done the labels.
> > > >
> > > >
> > > > --
> > > > Best regards,
> > > > Odhiambo WASHINGTON,
> > > > Nairobi,KE
> > > > +254733744121/+254722743223
> > > > _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
> > > > Damn!!
> > >
> > > > _______________________________________________
> > > > Testing mailing list
> > > > Testing at lists.pcbsd.org
> > > > http://lists.pcbsd.org/mailman/listinfo/testing
> > >
> > > Personally when I do a backup / clone to another drive, I prefer to
> just
> > > use "tar" to backup
> > > all the files individually, and then re-create the partitions on the
> new
> > > disk fresh, since geom may
> > > be different between them, etc.
> > >
> > > So, heres how I would do it:
> > >
> > > # cd /
> > > # tar cvjf /usr/myback.tbz --exclude ./usr/myback.tbz --exclude ./dev .
> > >
> > > Then format / mount the new disk somewhere, like /mnt, and extra the
> > > archive:
> > >
> > > # cd /mnt
> > > # tar xvpf /usr/myback.tbz
> > >
> > > Then if you partitioned correctly, you should be able to boot right off
> the
> > > new disk :)
> > >
> >
> >
> > Hi Chris,
> >
> > What is the disadvantage of using dump|restore after partitioning
> correctly?
>
> None that I'm aware of, I'm not a big user of 'dump/restore', I'm a 'tar'
> guy myself,
> so I was just chiming in with an alternative way to do the same thing :)
>
> >
> > Anyway, you did not answer my question, or I haven't understood your
> point.
> > I understand the disk geometry may differ, but I only use dump after
> > partitioning, in which case I believe only files are written over, and
> > nothing to do with geom, right?
>
> I believe that is the case, I was thinking of what you said about using
> "dd" though, which
> will try to grab everything, including partitioning.
>
> >
> > How can I duplicate my disk1 - the one I presented - and still ensure the
> > result I get has journaling enabled?
> > I am looking at a situation where, if disk one failed, I could simply
> > disconnect it, connect disk 2 into the same channel and boom!
> >
>
> Gotcha. It sounds like you want to use "gmirror" though, to keep a
> consistent copy of whats on disk 1, on disk 2, and do a failover, right?
> Otherwise you'll be stuck having to keep manually syncing data from disk1 ->
> disk2?
>

Maybe I should. Can you detail how I can do that with my current disk, or
point me to a doco? gmirror ufs+journal.
I actually didn't want to use gmirror. I'd like to manually sync the data.
Why? Because I intend to do the syncing on a weekly basis to give me the
possibility of recovering a file that I might have deleted. I am not sure
gmirror is good for this


>
> However if you want to go the manual route, then you should be able to do
> this with either tar/dump
> no problem. You'll just need to re-create disk2 with the same partition
> layout or similar to disk1,
> make sure you setup your single "/" partition with gjournal and you'll be
> in business.
>

That is what I wanted to do. The problem is that I am not able to read the
label information on the current disk. What am I missing?


>
>  The only weirdness I forsee is that we use "glabel" to setup
> /dev/label/rootfs0 as a link to /dev/ada0s1a.journal or whatever the
> particular devicename is you are using. If a failure occurs and you need to
> swap disks, you may need to manually label the new partition when you try to
> boot from it, otherwise you may get mount failures when it hits /etc/fstab
> :)
>

Now you hit the nail on the head! Is glabel usage the reason I am not
getting any meaningful output from disklabel?

Well, can I use glabel to

1. read the label info on the current disk1
2. write the label info on the disk 2 (both disks are 320GB)
.. then mount disk 2 on, say, /disk2 and dump thw whole of disk 1 into
disk2??

That's what I am after at the moment:)

-- 
Best regards,
Odhiambo WASHINGTON,
Nairobi,KE
+254733744121/+254722743223
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
Damn!!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.pcbsd.org/pipermail/testing/attachments/20101101/764e521f/attachment.html>


More information about the Testing mailing list