Bug#435834: installation-report: various problems with RAID1/cryptsetup/LVM
Hello!
At the end, I re-installed the system to check the reproducibility of
the bugs I encountered the first time. Sorry for the delay, the
re-installation process was finished last Sunday, but I didn't find
the time to report it before.
On Fri, 03 Aug 2007 16:04:31 +0200, Luca Capello wrote:
> Boot method: netboot
> Image version: etch-4.0r0 (current on the FTP site)
This time I booted from a etch-4.0r0 multiarch (i386/amd64/powerpc)
CD.
> 1) partition step
>
> I firstly created the /boot partition on both SATA drives:
> despite the fact that after the installation /dev/sdb1 is mounted
> as /boot and /dev/sda1 as /boot.bck, I remember to have set /boot
> on the first available device (port1/channel1 of the Promise
> TX4), thus it seems that the linux kernel missmap the Promise TX4
> devices [1]. Anyway,the remaining spaces was used for a RAID1
> array, then encrypted and finally LVM was added. After having
> partitioned the LVM space, I realised I forgot the swap space.
> But I haven't any more the "Configure LVM" option in partman.
This problem is still present: as soon as I configur the LVM and
finalize it (i.e. choosing the "Finish..." option in the LVM menu),
the "Configure LVM" option in partman disappears.
> 2) boot
>
> At reboot, the machine didn't boot: I tried to play with the
> motherboard boot options, but without success. I then detached
> both disks and attached them to another machine to be sure the
> installation process has installed the necessary files. /boot
> (/dev/sdb1) correctly had a grub/ folder and all the linux files,
> while /boot.bck (/dev/sda1) was empty. I then tried to clone the
> boot sector, still without success (and this is why both disks
> have a bootable partition). As a final test, I attached one disk
> at a time and the machine could boot with /dev/sdb attached to
> port1/channel1 of the Promise TX4 and /dev/sda attached to
> port2/channel2.
This is a bit strange: at reboot (without touching the SATA cables),
GRUB blocks after Stage1.5 (error, but this can be the old MBR from
the previous installation which I forgot to erase). So I swapped the
SATA cables and the machine boots, at least until error 3 below.
What's strange is that the disk which can boot was on port2/channel2
during installation and its first partition is /boot.bck (recognized
thanks to different labels), while the one that cannot boot is /boot
(and indeed it contains the correct files).
> 3) initrd
>
> The first boot stopped because vgsimba couldn't be find. So, I
> booted the etch-4.0r0 multiarch (i386/amd64/powerpc) rescue mode
> and manually set up the chroot (AFAIK there's no automatic way
> for an encrypted device). `update-initramfs -k all -u`
> complained about an error in /etc/crypttab and indeed
> /etc/crypptab was empty: I filled it with
>
> md0_crypt /dev/md0 none luks
>
> and at the next boot (after having created an updated initrd) the
> machine booted with no problem.
This is still needed.
Again, logs of the first and second installation are available upon
request.
Thx, bye,
Gismo / Luca
Reply to: