xfs
[Top] [All Lists]

Re: XFS - hard drive dying

To: A JM <vbtalent@xxxxxxxxx>
Subject: Re: XFS - hard drive dying
From: Iustin Pop <iusty@xxxxxxxxx>
Date: Sun, 4 Sep 2005 15:05:13 +0300
Cc: linux-xfs@xxxxxxxxxxx
In-reply-to: <a0f05e8905090404594af9f6d2@mail.gmail.com>
Mail-followup-to: A JM <vbtalent@xxxxxxxxx>, linux-xfs@xxxxxxxxxxx
References: <a0f05e8905090215183e30f40c@mail.gmail.com> <a0f05e8905090303584075a22e@mail.gmail.com> <4319B8DD.7080200@gmx.net> <a0f05e8905090311012f7b018e@mail.gmail.com> <20050903194320.GA6857@saytrin.hq.k1024.org> <a0f05e890509031318277a5425@mail.gmail.com> <20050904095639.GA8344@saytrin.hq.k1024.org> <a0f05e8905090404594af9f6d2@mail.gmail.com>
Sender: linux-xfs-bounce@xxxxxxxxxxx
User-agent: Mutt/1.5.10i
On Sun, Sep 04, 2005 at 07:59:05AM -0400, A JM wrote:
> The volume group I created was as far as I know a "default" setup...
> 
> If I'm going to try to re-create the VG, the /dev/hdb1 (dying 200gig Maxtor 
> ) will be dd' d to a 250gig drive, different size. Since I'm dd'ing the 
> drive and it's creating a partition /dev/hdb1 on the new drive will that 
> cause a problem? I'm guessing not, since it's essentially identical just 
> larger?
Well, a dd copy will have make /dev/hdb1 the same size as the old
/dev/hdd1, but since the /dev/hdb is larger it will have some space
unused.

> Just thinking out loud here, since I'm trying to fix (xfs_repair) a VG and 
> the essential data I'm looking for "should" still be intact on the new drive 
> I just dd'd from the dying one. How will the newly created VG on dev/hda4 
> know to incorporate the exisitng VG on the second drive /dev/hdb1? and will 
> it matter if the VG on /dev/hda4 is not identical to the original as long as 
> it knows to incorporate all of /dev/hdb1 into the VG and the VG name is the 
> same therefore I can run xfs_repair on it. 
NO NO NO, you must NOT create a new VG. You must re-activate your old
VG. Basically, LVM writes information about the volume groups onto each
physical volume. So, if you have /dev/hda4 INTACT and /dev/hdb1 at least
partially intact, you should be able to tell LVM to re-activate your old
VG.

I repeat - what do pvscan and pvs say? And what do vgscan and vgs say?

> Don't you have to set the "blocks" or something when incorporating a new 
> drive into the VG?
If you would re-create the VG, yes. But you trying to re-activate, not
re-create.

Regards,
Iustin Pop


<Prev in Thread] Current Thread [Next in Thread>