[plug] Data recovery with Gentoo system rescue cd

[caston at arach.net.au]

Quoting Craig Ringer <craig at postnewspapers.com.au>:

> On Fri, 2004-07-23 at 16:04, caston at arach.net.au wrote:
> 
> > I'm sure sure if there is an issue with these drives but I was using a 
> > hardcano 11 and felt that the drive temp was usually above what it should
> be. 
> > I watched it climb to 35degrees at the installfest.
> 
> 35 degrees seems pretty normal to me. That's where my RAID disks in my
> server at work sit normally, but they can climb to more than 45 degrees
> under heavy load. My home desktop's disks tend to be more like 25
> degrees, but then they're very well cooled and doing very little.
> 
> > A few questions:
> > 
> > Would it be worth using dd to dump the entire drive contents to the new
> drive 
> > or should I rebuild from scatch? 
> 
> dd will get you a more exact copy (down to the same inode numbers etc),
> but chances are that doesn't matter. When migrating data off failing
> disks I tend to use dd where possible because it's fast and easy.
> 
> If dd cannot read some sectors and fails, it is probably safer to make a
> new filesystem on the target and copy all the files instead ('cpio -p'
> is good for this) rather than using dd with conv=noerror.
> 

thanks Craig,

Right now I am doing cp -r * 

The source is ext3 and the target is ext2. I was in such a rush that I didn't 
adding journaling to the new disks file system.

What is the difference between this and cpio -p ?

I believe the last time I used cpio was unpacking oracle disk images and I 
didn't know what it was then and probably still don't now.

> > What is the dd command and block size I should use?
> 
> I find that bs=1M (or more) massively improves the transfer rate. 
> Otherwise, a normal
> 
> dd if=/dev/source_device of=/dev/target_device bs=1M 
>
Thanks.

I'm feeling that I may rebuild from scratch anyway just incase. 

Does anyone here know if KDE or anything else major is currently broken in sid?

> should do nicely. I would recommend building a new partition table on
> the new disk and doing partition-by-partition copies rather than trying
> to copy the entire disk, as I've seen some odd results with full disk
> copies in the past. 
> 
> You will need to reinstall the boot loader on the new disk, too.
> 
> Of course, with this sort of thing there is no one "right" way, and in
> the end so long as you get your data off safely it's good enough.
> 
> --
> Craig Ringer
> 
> _______________________________________________
> PLUG discussion list: plug at plug.linux.org.au
> http://mail.plug.linux.org.au/cgi-bin/mailman/listinfo/plug
> Committee e-mail: committee at plug.linux.org.au
>