Hi, W. Borgert writes: > Hello, > > today I sat down together with an experienced Linux/UNIX admin and > we went through the complete installation with RAID1. Unfortunately, > we didn't succeed. We have two identical HDs and created four > partitions on each in an identical layout. We sat all(!) partitions > to type 'FD'. We created the MDs and created three ext3 and one > swap (all on top of MD). We would like to have XFS, but it was not > in the menu, but that't not a big issue. Problems: > > 1. We were not able to create all MDs from the menu. At the > beginning, the first partition of the first SCSI disk did not > appear in the d-i menu. We created all with mdadm. > > 2. Deleting MDs from d-i didn't work. The MDs always showed up > again and again!
yapp, confirmed: menu doesn't, shell does (always?) work Another option is to zero out the superblocks, or at least the beginning of a partition with something along the lines of dd if=/dev/zero of=/dev/sdaXX bs=1024 count=64 do not take this command lightly. I did use it intentionally to zero out the complete disks with of=/dev/sda and no count ... it will run until "no space left on device". > > 3. The installation of the kernel failed completely with sth. like > /usr/sbin/mkinitrd: Cannot determine root device. well, this is a completely different game. I have not yet gotten through to this point. There was an interesting thread on debian-user lately ... //me searching, here we go: http://lists.debian.org/debian-user/2004/debian-user-200402/msg05104.html and references therein. As a possible workaround: install a small / partition elsewhere (no raid), reserve space for the read / partition on raid. Install. Now once that configuration is up, mount the raid /-partition on /new-root, say, copy / over to /new-root (cp -a), chroot /new-root, mount /proc and stuff in there. Now try to install a new kernel and see, how that goes in the chroot-ed environment. I have tried this, but not really come to a conclusion, of whether this works, should work or cannot work. Consider this a timeconsuming experiment. > We tried to disable the first MD (not possible from the menu, but > mdadm worked), and did / with ext3 and without RAID1, but error > (3.) still appeared. What's wrong with my setup? > > Thanks in advance for any help! > > Cheers, WB You asked: will this be in d-i at some point. The answer is: if someone creates the neccessary patches. But in more detail: - I think its safe to say, it will not be in beta3. - Keep in mind, that Pauls effort is not yet an official thing. - so IMHO it will be a while. I know that others are out there waiting for exactly this: have raid, lvm, devfs, xfs available immediately after installation. so, if we help it in: yes. Not to forget: "Thank you, Paul!" and whatever testing I can do to help, I'm willing to switch on my "noise machine" any other time ... :-) (OT: in case you have doubts about the unbelievable noise of a Seagate Barracuda --- get yourself some :) Cheers, Erich -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
