Bug#573762: shouldn't include mdadm into initramfs at all if INITRDSTAT='none'

[Marc Haber]

On Sun, Mar 14, 2010 at 06:58:25PM +0100, martin f krafft wrote:
> also sprach Marc Haber <mh+debian-bugs at zugschlus.de> [2010.03.13.1935 +0100]:
> > Why is mdadm included in an initrd for a system that doesn't use
> > any RAID in the first place?
> 
> Because it made more sense this way. Adding RAID to initramfs is
> hardly expensive or requires a lot of space, and if you accidentally
> set INITRDSTART=none, you'll be glad.

I actually missed a real error message in the warnings, which prompted
this bug report to be filed.

> > Please think about implementing an automatic or manual option that
> > prevents mdadm from going into the initramfs.
> 
> mdadm will eventually switch to using an event-based, incremental
> assembly algorithm (see the experimental branch). Then, mdadm will
> always be in the initramfs anyway.

I see.

> You always have the option to compile your own kernel,

I am doing so.

>  override the initramfs hook script,

How do I do this? Putting an "exit 0" script in
/etc/initramfs-tools/hooks/mdadm doesn't seem to prevent
/usr/share/initramfs-tools/hooks/mdadm from being executed.

>  or send patches that do what you want in a way that keeps mdadm as
>  generic and flexible as possible.

I think that overriding the hook script would be a fine solution if it
did work. Maybe I'm doing something wrong?

Greetings
Marc

-- 
-----------------------------------------------------------------------------
Marc Haber         | "I don't trust Computers. They | Mailadresse im Header
Mannheim, Germany  |  lose things."    Winona Ryder | Fon: *49 621 72739834
Nordisch by Nature |  How to make an American Quilt | Fax: *49 621 72739835