Bug#727708: systemd status when using multiple block device layers (MD/LVM/dm-crypt) below the root-fs

[Tollef Fog Heen <tfheen@err.no>]

]] Christoph Anton Mitterer 

> On Tue, 2013-12-31 at 21:07 +0100, Tollef Fog Heen wrote:
> > That's handled by the initramfs where we currently don't use systemd.
> > (It's supported upstream to do so and we might eventually investigate
> > that, but I don't believe anybody has done any work on it for Debian.)
> Sure... but
> - using systemd inside the initramfs _may_ come at a later point

We'll tackle any bugs if we decide to do that.  I don't see any point in
trying to figure out any bugs we might run into if we go down a route
nobody (in Debian, AFAIK) has even tried as an experiment.

> - even though the root-fs (and the resume-fs) is mounted in the
> initramfs image... it may still interfere with the boot process by
> systemd, e.g. when the later might accidentally try to re-setup such
> dm-crypt mappings or else... which were already set up in the initramfs.

Why would it do that?  Any such re-setup would be a bug, and I'm not
aware of any bugs in that respect.  Do you have any bugs where this
actully happens today?

> - all the questions, about whether systemd supports stacking of multiple
> block layers is also interesting for non-root-filesystems.

systemd mounts a block device.  How that block device is up to multiple
components such as lvm, cryptsetup, etc.  It's correct that systemd
currently does not support keyscripts, a deficency I was looking into
just last week.

-- 
Tollef Fog Heen
UNIX is user friendly, it's just picky about who its friends are