Hi, Cyril Brulebois <kibi@debian.org> (2023-05-03): > It's entirely untested for the time being, but I've tested a few things > inside an installer context, checking how hostname behaves when setting > and when getting the hostname (even if I already alluded to it in the > other bug report), and picking one of two possible options. You can find > details in the comments above the command getting added: > https://salsa.debian.org/installer-team/preseed/-/commit/2d83a156b3e50e3a35dd9964090967ac1761f85f > > I'll probably just find a dnsmasq config to tweak to force hostnames, > check a netboot/gtk/mini.iso against it, and upload if that behaves > better than D-I Bookworm RC 2. Setting “hostname=foo” at the very end of the ISOLINUX or GRUB prompts, respectively under KVM and its default configuration (no DHCP-provided hostnames) and on a Dell laptop connected to a dnsmasq assigning DHCP hostnames: - With RC 2, the hostname prompt is skipped, and “foo” is used in both cases. - With the modified netboot/gtk/mini.iso, the hostname prompt is skipped as well, “foo” is used under KVM, but “dellicious” is used on baremetal despite the hostname=foo trap. And when I say “is used”, I mean ends up being stored for cdebconf under netcfg/get_hostname (at least according to /var/lib/preseed/log), instead of “foo” previously. Notes: - I didn't actually perform any of those 4 installations in full. - Package just uploaded. - Daily builds dated 2023-05-04 should have all the required bits; I might trigger an out-of-crontab build for amd64/i386 later today if that helps people start testing right away. > In the meanwhile, if you spot something fishy in the reasoning or in > the implementation, please let me know. Still valid. Cheers, -- Cyril Brulebois (kibi@debian.org) <https://debamax.com/> D-I release manager -- Release team member -- Freelance Consultant
Attachment:
signature.asc
Description: PGP signature