[prev in list] [next in list] [prev in thread] [next in thread] 

List:       evms-devel
Subject:    Re: [Evms-devel] Remapping lost LEs of an LV to a new disk
From:       Kevin Corry <corryk () us ! ibm ! com>
Date:       2003-01-09 13:38:55
[Download RAW message or body]

On Wednesday 08 January 2003 20:39, Tuomas Jormola wrote:
> On Wed, 2003-01-08 at 16:09, Kevin Corry wrote:
> > On Wednesday 08 January 2003 08:35, you wrote:
> > > It wouldn't bring the data back but perhaps after this kind of
> > > operation I might be able to make the filesystem mount and get some
> > > data copied from it (and huge amounts of IO errors).
> >
> > There may be some things we could try to allow you to access the
> > filesystem data. Of course, this all depends on whether Reiser will mount
> > that filesystem at all, but maybe it's worth a shot.
> >
> > Currently, the kernel driver will simply return errors when it gets a
> > request for a section of the LV that is missing. I could send you a
> > kernel patch that would instead return success, and simply discard the
> > data for that section of the LV. That might be enough to trick the
> > filesystem into thinking there are no I/O errors.
> >
> > There may also be some manual tricks we can play with the metadata on the
> > new disk that you added to the group, but would be much more labor
> > intensive.
> >
> > Let me know if you are still interested in trying any of these ideas.
>
> Now this is something I call support :) Thanks, but fortunately I
> already managed to handle the situation with great success. I borrowed a
> big disk from a friend and copied as much data as I could from the
> partial block device inside the container using the great dd_rescue
> <URL:http://www.garloff.de/kurt/linux/ddrescue/>. It basically does the
> same as dd but it ignores hardware errors (usually bad blocks) and fills
> the data that cannot be read with 0x00 bytes. Now that I had the
> filesystem in one (~78GiB :) file there would be no I/O errors while
> reiserfsck tries to access and fix the filesystem. It took quite a long
> time to finish but it actually was able to fix the filesystem and then I
> simply mounted the now-resized-and-fixed filesystem using -o loop mount
> option. It worked great, and I was able to recover like 98% of the data
> that wasn't backed up and had some value.

Good to hear that you recovered your data! Your method was going to be one of 
my other suggestions. I know there are certain (obscure) flags to dd that 
accomplish the same thing that you did with ddrescue, by ignoring errors and 
returning zero-filled data. I am quite suprised that the filesystem was able 
to repair itself. Usually with that much of the volume missing, it is often a 
lost cause.

-- 
Kevin Corry
corryk@us.ibm.com
http://evms.sourceforge.net/


-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com
_______________________________________________
Evms-devel mailing list
Evms-devel@lists.sourceforge.net
To subscribe/unsubscribe, please visit:
https://lists.sourceforge.net/lists/listinfo/evms-devel
[prev in list] [next in list] [prev in thread] [next in thread]