Re: beta4 status
joeyh@debian.org (Joey Hess) writes:
> What we have right now in the sarge_d-i CDs and in unstable is very
> close to the final release. Only for powerpc and sparc are there
> possibly some changes in the initrds that are still building. For all
> the other architectures, there should be no changes, and so we need to
> do any final testing we can.
>
> Please test the images in unstable/main/installer-<arch>/current/ and in
> http://cdimage.debian.org/pub/cdimage-testing/sarge_d-i
Tried hppa and ia64. Results as follows.
cdimage-testing/sarge_d-i/hppa/current/sarge-hppa-netinst.iso
This fails to boot, because the palo command line is too long. The fixes I
made last week to d-i's config files along with the kernel udeb update to 0.54
wasn't sufficient, we also needed to update the debian-cd sarge/post-boot-hppa
to trim the palo command line. I have done so, and the change is in debian-cd
CVS. I'll be happy to try again as soon as we get a fresher image with this
fixed.
cdimage-testing/sarge_d-i/ia64/current/sarge-ia64-netinst.iso
The file overlap between libc6.1 and initscripts is still causing a fatal error
during the debootstrap run. I assume that means the fixed debootstrap wasn't
used, and/or one of the libc6.1 or initscripts packages is out of date? What
do we need to do to fix this?
Additionally, I note that the CDROM driver modules aren't loading right. On
a zx6000 I was able to shell out and modprobe both cdrom and ide-cd, at which
point CD detection worked and things progressed ok until debootstrap failed.
I've studied the cdrom-detect source, but it looks like it only tries to load
the 'cdrom' module and that only after it fails to find anything and interacts
with the user. What do I need to change to get the likely modules loaded
before cdrom-detect's postinst does its thing? I'd like to fix this for beta4.
And finally, ia64 systems use EFI for firmware, and it needs a FAT partition
to hold the elilo bootloader, etc. If I use the manual partitioning to create
one, but don't give it a mount point (in Debian we don't mount this partition
routinely), I get a complaint about the lack of a specified mount point and the
text would confuse an inexperienced ia64 user. How should I address this? Do
I need to add a partman-elilo method, or is there some other hack I can use to
keep the install process from complaining about a FAT partition not having a
mount point assigned? This qualifies as errata-only for beta4.
Finally, automatic partitioning for both hppa and ia64 is, of course, useless
until #244736 is addressed after beta4.
Bdale
Reply to: