Re: Booting Debian/testing fails

[hendrik@topoi.pooq.com]

On Sat, Feb 03, 2007 at 06:28:45PM -0500, Douglas Allan Tutty wrote:
> On Sat, Feb 03, 2007 at 05:44:40PM -0500, hendrik@topoi.pooq.com wrote:
> > On Sat, Feb 03, 2007 at 08:50:27AM -0500, Douglas Allan Tutty wrote:
> > > On Sat, Feb 03, 2007 at 08:03:02AM -0500, hendrik@topoi.pooq.com wrote:
> > > > On Sat, Feb 03, 2007 at 12:57:09AM -0500, Kevin Mark wrote:
>  > > 
> > > > For example, where do you find details on why rescue mode, swapped hard 
> > > > drives (/dev/hda <-> /dev/hdc) when I asked it to start a shell in the 
> > > > context of my root partition but not when I asked it to start a shell in 
> > > > the installer context?  In fact running fdisk /dev/hda in the root 
> > > > context showed me a perfect partition table for /dev/hdc, except that 
> > > > all the partitions were labelled as being on /dev/hda.
> > > > 
> > > > Now I know the boot-loaders have provisions for swapping hard-drive 
> > > > letters.  But why were they invoked?
> > > > 
> > > > This is the kind of detail that needs to be documented.  And access to 
> > > > wource code is no longer a solution, even for experienced programmers -- 
> > > > there's just too much undocumented context for each piece of the 
> > > > hundred-million-odd lines of code that constitute Debian that that's 
> > > > only practical for specialists in the particular subsystem under 
> > > > investigation.
> 
> > > 
> > > Do you need to know the _why_ of that, or would you have liked a
> > > heads-up and what to do about it?
> > 
> > I needed to know what to do about it.  Not knowing the why, I decided I 
> > couldn't trust the rescue system and improvised a workaround that 
> > involved installing a new Debian system on a spare partition on 
> > /dev/hda2 (fortunately I still had some space) and using *its* lilo to 
> > establish bootability of /dev/hdc3.
> > 
>  
> > I know that Debian has ways of futzing the BIOS so that drive letters 
> > are different from the standard ones.  What I don't know is why it 
> > decided to do this as a rescue attempt.  My lack of understanding made 
> > the rescue system untrustable, hence unusable.  And I'm not at all sure 
> > that posting the details on, say, debian-user would have resulted in an 
> > answer sufficiently authoritative to be trustworthy.
> > 
> Hi hendrik
> 
> I would have called it a bug in the installer and submitted it.  It
> would end up on debian-boot where a trustworthy reply should have been
> available.
> 
> If you have a chance, make a current daily-build netinst.iso an see if
> the problem still exists.  If it doesn, file a bug.

There may be reasons for it -- for example, the installer might examine 
the files within the partition and decide that that partition was meant 
to be booted with the drive letters swapped.  I once has a Mandrake 
system that insisted on relettering its drives in lilo so that hda was 
hde.  I have no reason why.

-- hendrik