[Yaird-devel] Bug#351183: --run does not fix the disk renaming problem

[Jason Lunz]

On Tue, Jun 06, 2006 at 02:55:50PM -0700, dean gaudet wrote:
> !case "`mdadm --version 2>&1`" in
> !        mdadm\ -\ v1.[1-9][^0-9]*)
> !                mdadm --assemble <TMPL_VAR NAME=target> --uuid <TMPL_VAR NAME=uuid> \
> !                        <TMPL_LOOP NAME=components> <TMPL_VAR NAME=dev></TMPL_LOOP>
> !                ;;
> !        *)
> !                mdadm -Ac partitions <TMPL_VAR NAME=target> --uuid <TMPL_VAR NAME=uuid>
> !                ;;
> !esac
> 
> hmm well i need to handle mdadm v0.x too i guess... i'm trying to avoid 
> added grep to the initrd.
> 
> let me know if the above is generally what you're looking for -- an 
> alternative is to test the mdadm version at mkinitrd time but i didn't see 
> a way to do that without mucking in the perl (which i don't mind doing 
> either).

remember, --run *does* work for older versions of mdadm. It just doesn't
handle the drive reordering case. So we'd want to add the --run in the
non-Ac-partitions case if we go with this approach.

> i probably won't be able to test a patch for a week or two though.

I've got some machines running sarge with md root partitions and bpo
kernels with yaird initramfses. I don't think I can do any
upgrade-to-etch testing, but I can generate some initrds and try them
out on various drive loss/reorder scenarios.

Jason