David Nusinow wrote: > Hi Sam, > > On Mon, Feb 09, 2004 at 06:50:16PM -0500, Sam Newman wrote: > > i downloaded the latest sarge (02-08-2004) netinstall image to install > > debian on a test box to see how i like it. everything runs great until i > > get to the lilo part. it asks me where to install, but i'm not too sure > > what to do. i tried installing it to just /dev/hda but that didn't work. > > this is going to be the only operating system on it, so there aren't any > > windows conflicts. also, the default install path for lilo doesn't work. > > > > i also tried to install the grub boot loader and it froze at 80% (using the > > default hda0 option). > > This is a known bug right now. See #231626 all the bugs it is merged with. I think I understand this bug. It goes something like this: 1. User downloads the non sid_d-i version of a netinst CD, ignoring the warning that it won't work and the encouragements to use sid_d-i instead. 2. The initrd has rootskel 0.61 on it. This version of rootskel contains an apt-install that depends on changes made in version 0.051 of base-installer to work. 3. Since this is not a sid_d-i CD, it contains base-installer 0.046 from sarge. (Note that base-installer 0.047 is also present in unstable for the arm and sparc architectures.) 4. apt-install works ok until base-installer is run. After that, it will want to actually install stuff (like the kernel), rather than queuing it. Since the new method it uses to run apt depends on base-installer having bind mounted the CD into /target, this will fail. 4. Later the boot loader fails to install as there is no kernel. There might be a different chain of failures when a CD is not involved in the install, and users have a too old or too new initrd vs the default distribution anna downloads d-i from (sarge). FWIW, I did two installs today on two different machines, using yesterday's sid_d-i netinst image, (one USB, one standard vmware), and both worked fine. -- see shy jo
Attachment:
signature.asc
Description: Digital signature