Re: [Sheflug] Hard disc failiure on a reiser file system.

[Seb James <seb [at] xxxxxxxxxxxxxxxxxxxxxx>]

On Tue, 2004-08-10 at 19:58, Rob Keeling wrote:
> ----- Original Message ----- 
> From: "Alex Hudson" <home [at] alexhudson.com>
> To: <shef-lug [at] list.sheflug.org.uk>
> Sent: Tuesday, August 10, 2004 7:42 PM
> Subject: Re: [Sheflug] Hard disc failiure on a reiser file system.
> 
> 
> > On Tue, 2004-08-10 at 19:35 +0100, Rob Keeling wrote:
> > > I have started to see messages like :-
> > >
> > > Aug 10 19:24:51 mail kernel: hda: dma_intr: error=0x40
> > > { UncorrectableError }, LBAsect=2421646, high=0, low=2421646,
> > > sector=365320
> > >
> > > My guess is that things have cooked temp wise, but the question is,
> > > whats the best way of recovering from this,
> >
> > Backup the data immediately and replace the HD. I also wouldn't turn off
> > the machine until you're sure you've got your data off it; the HD might
> > not come back ;)
> >
> Thinking of taring the / folder, pipeing via ssh to a box with spare space
> before shutting down
> the machine overnight to see if things improve tomorrow!
> 
> guessing I need to skip /proc but thinking of something like
> 
> tar cxvO --ignore-failed-read --exclude /proc | ssh remote "cat
> >/backup/mailbackup.tar"
> 
> would mean I could re-install suse 8.2 and then extract the tar file to put
> everything back.
> 
> Any thoughts?

This is how I routinely copy systems about: 

On my mirror machine, I run sshd and have rsync installed. In the
/mirror directory I'll create a directory as the destination. In this
case /mirror/computer_name. On the computer "computer_name" I then run:

/usr/bin/rsync -az -e ssh --delete --exclude=/proc --exclude=/sys
--exclude=/dev / 192.168.0.1:"/mirror/computer_name"

Voila, whole system copied (well, you may wish to go off and read the
paper at this point...). The command above will also of course do an
incremental update and is what I use for automated nightly backups.

Now, boot "computer_name" with some rescue cd or another (I use a gentoo
install cd) having swapped hdds around. mkfs and mount your new
partitions and then a similar rsync command will pull everything back
again. Couple of small tweaks (make sure /proc, /dev, /sys directories
are there and that /etc/fstab is correct for your new hdd) then run grub
and reboot.

This is the single most brilliant thing about unix systems - the ease
with which you can move them around; it's just a bunch of files, after
all!

Seb

___________________________________________________________________

Sheffield Linux User's Group -
http://www.sheflug.co.uk/mailfaq.html

  GNU the choice of a complete generation.