Re: About the continuity of SlugOS/LE + Debian
José Antonio Insua wrote:
> Some time ago I installed in my NSLU2 using SlugOS/LE and I bootstrapped to
> That was a neat solution. I was able to debug whatever problem I had ( and I
> had some ) from the SlugOS and after correcting them, boot my Debian.
> From a user point of view, the minimal SlugOS system is very handy to debug
> any problem or to make a fsck to the disk filesystem (without needing to add
> a serial port).
Yes, it would be great for us to work out how to duplicate that useful
recovery filesystem functionality on Debian embedded systems.
> Now, I would like to have a 2.6.18 debian kernel, with the new open source
> network driver and the greatest patches. One way to achieve that is the
> standard debian install, but that will erase the SlugOS.
> Other solution could be awaiting for a new SlugOS/LE version, but as far I can
> see the bootstrap method is not recommended anymore and I would expect that
> the new OpenSlug kernel versions can be incompatible with the debian disk
The latest SlugOS/LE and Debian/NLSU2 kernels are pretty much identical
(as long as you have the correct kernel modules installed) from a rootfs
point of view.
> I can see three possible solutions:
> - Perhaps the SlugOS is not really needed, I don't fully understand the
> system and that's why I need the small SlugOS.
> - I can recompile ( and modify if needed ) myself newer versions of OpenSlug
> to work like the old SlugOS/LE, but because I see nobody doing that, perhaps
> this is a bad idea or this is too much work.
Bootstrapping Debian from SlugOS/LE is no longer supported (it still
works fine, but is not supported) since the Debian installer now
replaces that functionality.
> - And the last solution ( a deadly elegant one IMHO ) is to have an OpenSlug
> distribution with kexec support (recently added to 2.6.21) that would be
> functional and would be able to load a debian kernel from the hard disk. This
> would be great to test software, new installations, etc ...
SlugOS/LE alpha versions already have that functionality, and I expect
that the next Debian/NSLU2 kernel will have it too.
> If somebody is working on the solution I would be glad to know that, and if it
> is not done, I would also like to know why. If the reason is lack of
> manpower, perhaps I could take that as a personal project :) .
The best thing would be to work out how to add a small recovery
filesystem to the debian installer, to which the boot process falls back
to if it cannot load the rootfs from the hard disk.
> Kind regards to every one in this great project :)
-- Rod Whitby
-- NSLU2-Linux Project Lead