launchpad and fai for ubuntu 8.10 (both server and client)

[Russ Allbery]

"Adrian Gibanel Lopez" <agibanel at alumnes.udl.cat> writes:

> No. I've already said something about the reason for not doing so. I do
> not know which it is the actual bug.
>
> Is it busybox's mount command not being updated to the --bind and
> --remount syntax?
>
> Is it initramfs-tools and live-initramfs not using --bind --remount and
> -o bind -o remount syntax in the right places (Before and after
> gathering a dinamic mount which supports --bind and --remount syntax?

I would file two bugs, one for busybox not supporting the new syntax and
another for initramfs-tools using the old syntax with a mount binary that
doesn't like it.  In both cases, reference the other bug so that people
can see the situation.  But I would get the bugs filed, even if it's not
clear where it should be fixed, since that will hopefully start the
process of converging on a good solution.

> I do not want to bother other packages till someone repeat my debugging
> process in order to make work a vanilla Ubuntu 8.10 fai package to
> server Ubuntu 8.10 to the install clients.

That's laudable but I think too conservative in practice.  Bugs don't hurt
anything, and I think you've been quite thorough.  You've certainly done
way more testing and research than go into most bug reports.

I'd like to duplicate your work, but unfortunately I don't have the time
to do so just at the moment.  I may be able to find the time over the next
few weeks, but the more people who are aware of the issues, the better.

> If you think that FAI-only developers as the ones who might be
> subscribed to this mailing list are not enough/sufficient to track the
> problem, well, we can discuss asking the other packages' developers.

I think it's fairly clear from your messages and analysis that there are
problems here that can't be fixed in FAI itself, which means that the
developers of the other packages will need to be involved sooner or later.

> And let's not forget my ldconfig.real call, in my opinnion it is a
> bug,...  but someone will have to track which mount call does the bad
> job, if it is a dinamic mount (What's the dinamic mount's package ) or
> an static mount (busybox). Or is it maybe a ld bug?

As I recall, this was the problem that the ld.so.cache file was out of
date and hence the system wasn't seeing newly available shared libraries?
Or am I misunderstanding the problem?

-- 
Russ Allbery (rra at debian.org)               <http://www.eyrie.org/~eagle/>