[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: d-i experiences

Am Di, 2003-09-16 um 01.58 schrieb Daniel Silverstone:

> I fired up a bochs machine and booted the netinst CD
> It sat looping saying "unable to load vesafb.o"

Fixed in cvs.

> Firstly, the framebuffer console is incredibly slow. But I can cope with
> this...

This is a problem of vmware. It is a lot faster on a "real" system.

> The first menu d-i presented me with having booted was in English. It's
> very very lucky I speak english, or I wouldn't have known to press Enter
> in order to be able to select English as my language of choice for the
> installer.

This will be dealt with, when we are ready to release. It is very esay
to rearrange priorities, so that the first thing you'll get to see is
the language dialog. However for the time being it is left like this for
debugging purposes.

> I selected 'detect cdrom' and d-i kindly told me there'd been an error
> whose text was 'EXIT' before succeeding to detect the cdrom

This is a known bug, although I do not remember which package is
responsible. I believe it is busybox, but I can't find a bug report.
Will investigate tomorrow.

> On the loading modules page, many of the descriptions don't fit, this is
> poor UI design

Reported as #209273. Should be easy to fix.

> On the second 'go back' I got an error. Apparently anna's postinst
> (whoever anna is) exited with status 2560 (whatever that means)

This is recently being discussed on this mailing list (#209343 and
#210967). There is a solution coming up.

> I decided to configure my NIC by DHCP. I selected this option and
> answered the questions it asked me. It then presented me with what
> looked like a post-DHCP configuration output and asked me if it was
> okay.

Yeah, work is needed there. Too many useless questions indeed.

> All I knew was that d-i had highlighted the PPP/serial option
> suggesting to me that DHCP had failed. (It hadn't)

PPP/serial should not be default at that point. This bug is perhaps
fixed with a recent main-menu checkin.

> I then decided to configure which mirror to install Debian from. But the
> UK appeared to be missing from the list,

Please file a bug about this.

> Having partitioned my drive the hard way I tried to create filesystems.
> The d-i businesscard installer appears to be missing reiserfs as a
> default option for filesystems. This is poor and should be rectified.

Reiserfs is as good as supported. We only need a module which provides
the kernel-modules for reiserfs. I wrote Herbert Xu a mail about this a
week ago, but got no reply :-( If the kernel module is loaded, partconf
will detect it and offer to format with reiserfs.

> This part of d-i failed to take notice of my partition types and kindly
> offered the option of creating swap on my / partition, and indeed
> creating an ext3 filesystem on my swap partition.

I too found that strange. But some people argued that partition types
are not more than cosmetic. I do not know about this though...

> As I fought my way through the partitioning and formatting options I got
> many error dialogs containing what look like the output of cfdisk, only
> containing piles of escape strings too.

Could you elaborate on this more please?

> I happened to chance upon the syslog terminal which seemed to be
> logging a whole pile of errors of the form:
> installer[21577]: WF: Retrying failed download of %s

This one is being tracked down at the moment.

> After installing base, the 'debian-installer' process exited and was
> restarted according to the syslog console

This is a new one. A bug report with more information would be

> I chose install-kernel, unfortunately d-i seemed to think I should
> install base again. It chundered through for a bit and then gave up,
> giving me the error that debootstrap had failed and I should look in
> /target/var/log/debootstrap.log for information.

Debootstrap is quite ignorant. It does not work if it finds files where
it wants to copy new ones. So a second invocation of base-installer
always fails due to failing debootstrap.

> If this is the current state of the installer, I'm very concerned that
> we'll not be able to release sarge on time; or that we'll have to fall
> back to boot-floppies for the most part.

As you see we are aware of most of the problems. Unfortunatly we are
lacking manpower to tackle all these problems. So any help is

PGP-Key: http://www.mmweg.rwth-aachen.de/~sebastian.ley/public.key
Fingerprint: A46A 753F AEDC 2C01 BE6E  F6DB 97E0 3309 9FD6 E3E6

Attachment: signature.asc
Description: Dies ist ein digital signierter Nachrichtenteil

Reply to: