[rear-users] Restoring mdadm system to a single-disk destination fails

Jeroen Hoekx jeroen.hoekx at hamok.be
Tue Jul 24 17:04:36 CEST 2012

On 24 July 2012 16:37, Dag Wieers <dag at wieers.com> wrote:

> In your case I would expect the following procedure:
>   - boot rescue image
>   - modify disklayout.conf to remove the md devices, and replace by sd devices
>   - rear recover
>   - remove the remaining mdadm stuff (this should be automated in Rear)
>   - reboot
> If we want to make it possible to have the interactive process to allow
> for mapping md devices to sd devices, we will have to analyse what is
> needed for this and Jeroen will have to give a green light.

I once started such an unRAID script. So it's not something I really
oppose. I stopped it because of time constraints and because I figured
it would not be used enough to warrant the effort.

> If we want to make it possible to allow these mappings before making the
> rescue image, we have to analyse what would be needed for this and Jeroen
> will have to give a green light.

This looks like the best way to me at the moment:

A new variable array SKIP_COMPONENT should be defined that skips that
component and moves all child components higher up the tree.

Just adding the mapping resulting from that variable to the mappings
file in /etc would  enable textual replacement in config files.

We have a generic solution to the problem in that case.

For mdraid in particular, we could add a script that detected if we
could not recreate an array because of missing disks and prompt the
user for the replacement.



More information about the rear-users mailing list