(No need to CC me.) On Sunday 06 November 2005 15:26, Marco Amadori wrote: > Alle 14:37, domenica 6 novembre 2005, Frans Pop ha scritto: > > Actually, I'm not sure where this should go. You can also make a case > > that it should go into hw-detect's disk-detect. > > I would go on with this idea too. Hmm. Then why does your proposal below implement it in partman? > > I guess it depends on how the presence of dmraid can be determined: > > - does dmraid need to be loaded for that? > > 2 things for array detection: modprobe dm-mod and "dmraid -s -c" , see > my previous post. > > > - does it need to be loaded for all systems (would be bad) or can a > > kind of preselection be made based on controllers that support it? > > If the dmraid invoking gives us "No RAID disks" we could simply skip. That still requires dm-mod to be loaded, right? I should like to avoid modprobing dm-mod for systems that do not have any of the controllers supporting dmraid. This is especially important for older systems using lowmem installations. > > Maybe discover could be used/extended for this? > > We should ask discover maintainers to add a dmraid call if it > finds /sbin/dmraid ? No, just to answer the question "does controller support dmraid". > 2 We ask in high priority to enable it and in medium debconf we enable > without asking. You are confusing priorities here: - critical: least questions asked - high: default mode - medium: very nice mode for more control - low: "expert" mode, all questions and messages shown So it should be enabled by default at high priority and ask the question at medium. > 3 Invoke partman on /dev/mapper/nvidia_jbdaccab and let the user > proceed as usual. How do you "hide" the physical devices? Or is that done automatically? Other d-i components that probably will need to be adapted are grub-installer and lilo-installer, but that can come after the changes in either disk-detect and/or partman.
Attachment:
pgprqA7o8pFdz.pgp
Description: PGP signature