Re: potato hard disk install bug
David West <email@example.com> writes:
> I'm using the Oct 14 materials in http://auric.debian/org/~aph/bf/
> in particular the 1.2MB boot floppies and base2_2.tgz.
> After problems getting the install script to find the kernel
> and drivers (I have them, but they apparently are not found,
> even after I create some symlinks which satisfy what the install
> script says it wants), I copied the debian kernel to my bootable
> hard disk's Slackware partition, rebooted into Slackware, set
> up lilo.conf to boot the debian kernel mounting the new debian
> partition as root, reran lilo, and rebooted, choosing the debian
> installation when lilo ran.
> All went well until after the message
> Starting internet superserver indetd
> Then it said
> Template does not contain a template: line at /usr/lib/perl5/Debian/DebConf/Template.pm
> line 66 <TEMPLATE_IN> chunk 1371
> INIT: Id "1" respawning too fast: disabled for 5 minutes
> At this point I gave up.
Well, sure. You're trying to run the installation materials on a
Basically, as I read it, what you are trying to do is not directly
supported. Obviously, if you use the standard "install from floppies
onto a clean root partition", then everything should work (or a big
bug should be filed).
> Remarks and questions:
> Given that I have base2_2.tgz and a compatible set of kernel+drivers,
> a working Slackware installation on another partition, and a working knowledge
> of lilo, the only reason I started by booting from the floppies is
> a suspicion that just untarring base2_2.tgz into an empty partition
> is not enough to create a "correct" minimal install, i.e. that
> dbootstrap may need to be run to do something that I didn't know
> needed to be done. Can anyone give definitive info on this point?
You are correct. There are ways to take the base_*.tgz stuff and
make a working base system out of it; I haven't ever done it but it
should work given that you have the kernel, and modules, and some
config files in place.
> The debian install doc says, under hard disk install, says the
> boot floppies are not actually necessary, and that "any other boot method"
> can be used. I think the doc needs to be more explicit on this point, and
> also on the issue of exactly how much directory structure is necessary
> to get the installable files recognized.
Probably. More explicit comments would of course be appreciated,
probably in the form of a bug report agianst boot-floppies.
> I really wanted to install the idepci kernel instead of the larger default
> kernel, but the floppy install didn't seem to offer this as an option.
> (For the manual install attempt, I used the default kernel just to be on the
> safe side.)
That's because to install the idepci kernel, just use the idepci boot
and root and modules filesystems.
> I would guess that all of the above indicates that hard-disk installs are so
> infrequent that the relevant parts of the install scripts are much less well
> debugged than is the case for, e.g., CD installs (I'm only doing this because
> I don't have a CD drive), just as 1.2M floppy installs are so infrequent that
> no-one (apparently) tried one before potato was released.
It's possible, but I have done installs this way, installing all from
materials on a DOS filesystem, and it worked fine.
Bug reports appreciated. Without detailed bug reports (and this
report here was not quite detailed enough for us to try to reproduce
what you're doing) we can't make things better.
.....Adam Di Carlo....adam@onShore.com.....<URL:http://www.onShore.com/>