[ale] initramfs capable of a multi-device btrfs / new initramfs with dracut help?

[Michael B. Trausch]

On Fri, 2010-07-16 at 11:34 -0400, Brian W. Neu wrote:
> btrfs has RAID levels 0,1,10 built-in (and will have 5 & 6 some day in
> the probably far-off distant future)
> 
> I don't want to use mdraid because not only is it unnecessary but it
> prevents checksums from being able to identify a faulty drive.  mdraid
> really offers no integrity checking, which I'm not willing to do
> without anymore. 

If btrfs is seriously trying to give ZFS a run for its money, it's going
to adopt RAID5/6 and ZFS's non-standard RAID-Z, and hopefully sooner
rather than later.  I am also anxiously awaiting the data deduplication
feature, when it arrives.

However, it's going to be a while before I really trust btrfs.
Development on it has slowed way down, it would appear, which stinks.  I
love the speed of the FS, and I love the fact that it can dynamically
grow and shrink.  I like the whole notion of copy-on-write snapshots at
the file level, because that makes experimenting with very large files
like virtual machine images quite easy to do in an efficient manner
without having application-level support for stacking images.

It would be nice, though, if I could get rid of the mdraid setup that I
have currently.  I really am finding that I hate Linux's software RAID.
I have been trying for weeks to make it act intelligently, but it
doesn't seem to be really capable of that without lots of manual help.
I'm probably missing something (and honestly, I *seriously* hope that I
am) but as it is, I cannot reboot my server running the RAID without
physically being there to bring up the RAID.  That's not cool.  I had to
disable the automatic feature because it would always assemble the RAID
improperly and force a three-day rebuild process, which is even less
acceptable IMHO.

To be able to replace that with a filesystem that can intelligently
rebuild only the necessary data when a drive fails?  That would be a
delight!

	--- Mike